#foswiki 2010-12-08,Wed

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
***Bamieater has quit IRC (Ping timeout: 264 seconds) [00:05]
.... (idle for 18mn)
ArthurClemens has quit IRC (Quit: Leaving...) [00:23]
toffe82 has left [00:31]
...... (idle for 29mn)
Bamieater has joined #foswiki [01:00]
.... (idle for 18mn)
Bamie_ has joined #foswiki
Bamieater has quit IRC (Read error: Connection reset by peer)
Bamie_ is now known as Bamieater
[01:18]
Bamieater has quit IRC (Ping timeout: 265 seconds)
Bamieater has joined #foswiki
[01:23]
...... (idle for 28mn)
favioflamingo has joined #foswiki
foswiki_irc6 has joined #foswiki
[01:52]
foswiki_irc6hi i have issue with smtp
am installed 1.0.10 on windows / apache/ strawberry
setup smtp settings an it fails to send the email
also i get a lot of 'the connection was reset' errors
i set up using http://localhost/foswiki/bin/configure script
i am experienced twiki sys admin
[01:56]
.... (idle for 17mn)
pharveyare there any messages in the log? data/warn201011.txt or apache [02:15]
***MartinCleaver has joined #foswiki
MartinCleaver has quit IRC (Changing host)
MartinCleaver has joined #foswiki
[02:15]
foswiki_irc6thanks for replying , I fixed user registration by disabling smtp in configure , will check log now...
| 2010-12-08T01:48:19Z warning | SMTP auth: 500: Need MIME::Base64 and Authen::SASL todo auth - Trying to send without authentication | | 2010-12-08T01:48:20Z warning | ERROR: Can't send mail using Net::SMTP. <cpc3-nmal16-2-0-cust111.croy.cable.virginmedia.com[94.172.8.112]>: Client host rejected: Access denied at C:/strawberry/perl/lib/CGI/Carp.pm line 354
issue appears to be no sendmail on win 7 or in the strawberry win install
is ok until i want to reset user pw , then it needs smtp
[02:15]
pharveyThis sounds familiar. I think you need Authen::SASL if your SMTP requires auth
so you are missing a library
Foswiki should fall-back to Net::SMTP if it can't find the configured sendmail program
[02:18]
foswiki_irc6aha , ok
so if i install perl lib SASL this may fix?
[02:19]
pharveyI think so [02:22]
favioflamingohey, morning. for meta data, what significance does preference "type" have?
i can see name, and value
[02:27]
pharveyfavioflamingo: META:PREFERENCE needs to say if the 'type' is Local or Set [02:34]
favioflamingoah
ok
[02:34]
foswiki_irc6hey that perl module fixed it , big thanks !! , lastly, how can i delete users? [02:35]
pharveyfoswiki_irc6: delete 'retired' (non-current) users, or spam accounts?
Usually you don't want to delete the user topic of a non-current user; otherwise a new (different) user with the same name might be registered year(s) later, and inherit any access control privileges they had. Not to mention, making topic histories point to the wrong individual
so on my wiki, we just change the VIEW and CHANGE permissions so that only AdminGroup sees them
if you are using the "out of the box" login/user/password managers, I'm afraid you have delete their line out of the .htpasswd file manually
HTTPDUserAdminContrib IIRC was an attempt to streamline this stuff to make it more admin-friendly, I'm not sure how finished it is (ping SvenDowideit)
[02:36]
foswiki_irc6ok i will follow your advice on view an change perms , thanks [02:40]
pharveymany corporate installations just lean on the company LDAP server for user accounts, so they would use their normal user account management tool [02:40]
foswiki_irc6yeah thats how I have setup Twiki [02:40]
pharveyanyway, don't forget to delete the line from .htpasswd, otherwise the account will still work :)
(and if you have time, give HTTPDUserAdminContrib a try)
[02:40]
foswiki_irc6but this is just a noddy little install for me an maybe 1-2 other ppl i created 2 test users just wanted to delete them , im setting it up to make a basic document repository mainly for pdf attachments (help me go paperless office) [02:42]
pharveyah ok, no problem
favioflamingo: have you read http://trunk.foswiki.org/Development/QueryAcrossTopicRevisions ? CDot checked in some code, but I don't think it works properly yet.. anyway, just wondering if your SQL schema will handle querying against past revisions
[02:42]
favioflamingoya, it should, i changed it up a bit
ill have to read up on that first
[02:43]
pharveyanother thing you could do, if you just want your SQL database to stay synchronised with the .txt topics, is to do a store listener. DBIStoreContrib has store listeners which get called any time a topic changes; it's a more friendly (passive) way to listen for data change than overriding Foswiki::Meta
I think MongoDBPlugin overrides Foswiki::Meta though
[02:44]
favioflamingomy objective is to eliminate the use of txt completely tho, so... [02:45]
SvenDowideitHTTPDUserAdminContrib is 2 things
first up a mapper that uses other backend stores
and then i started to make user admin UI's - i can't recal how far i got
though the 1.1 groups UI and code is derived from that work
one big thing it does, is use the same database layout that apache's db_auth modules do
allowing more speed&portability
[02:46]
favioflamingogeneral question about meta preferences: are those values needed during query searching, or page rendering, split between the two? [02:48]
SvenDowideitand even uses apache's groups system
yes to both
[02:48]
...... (idle for 25mn)
_Dave_hey Sven [03:14]
SvenDowideitheya :) [03:14]
_Dave_have you thought about using an environment variable to get foswiki it's setlib.cfg path? that should pass through all mechanisms to execute perl at the end of a webserver? [03:15]
SvenDowideitpersonally, i detest the use of ENV vars
as it relies on a non-versioned something or other that could be quietly changed by anyone
(i have my cfg files versioned - that way i have a pretty good idea of what gets run)
i know other's like to use them for configuration management - both in running services and when building
but i've watched waytoo many people think they know what's happening in their system, only hours later to discover that some ENV var was not what they thought it was
[03:15]
_Dave_well it's the best guaranteed mechanism I know to pass a scripted path to a scripting system
forgetting to set it isn't the software's fault ;)
[03:18]
SvenDowideitlet alone sent releases to customers that can't be re-produced, because some **** built using a manually sent ENV [03:18]
gac410_Dave_ I think ENV is probably last resort ... do all hosted env's have the ability to set ENV variables? [03:19]
SvenDowideitwhich is why i avoid using ENV whenever i can [03:19]
_Dave_all of them I have worked on, you needed one to set things up right
but gac410, that's a good point...no one person has worked everywhere :D
however, I know hosting providers that would not let you run a perl app that forces "." at the front of @INC
[03:19]
SvenDowideityup, which really should not be needed - as you know the path [03:21]
gac410One interesting warning Lavr found. FindBin does it's magic in a BEGIN block, and will only work for the first use in a Persistent perl env. So it is strongly recommended to not use FindBin in Mod_perl, etc. [03:21]
_Dave_well you already have something that writes #!
why not make it also write a $my_library_is_at global?
that's in fact what I do with most of my perl apps
personally, I think depending on an ENV is far better than "." as the first component of @INC
but I'd rather not pick between two perceived evils, even if I don't perceive one as evil :D
[03:21]
gac410That's what setlib.cfg does. The problem is finding setlib.cfg in the first place. It's in the script dir. Which suexec hides from perl [03:23]
SvenDowideitmmm, better, but only marginally? [03:23]
gac410If we could find setlib, then . would not be needed. nothing else in Foswiki needs the . [03:23]
SvenDowideittbh, its all a bit mad [03:24]
_Dave_I agree lol [03:24]
SvenDowideitwe don't need more than one script
er, 2
configure, and the foswiki one
[03:24]
_Dave_but yeah, this is why I gave up years ago and now I have a module and a ./Configure script who's job it is to go through and put my own BEGIN block at the head of every executable script [03:24]
SvenDowideitand if confirure where to wrote the other
we'd be much better off
[03:24]
gac410er configure, unauthed and authed? [03:24]
SvenDowideitna, i think i've found enough docco that shows you can use 2 /Locations to point to the same real file
and have one Location be authed and the otehr not
don't have time to persur it yet, but..
[03:25]
gac410LocationMatch is pretty fussy about where it goes. and has some restrictions about what you can specify as auth. [03:26]
SvenDowideitor another eg using rewrite's
and i imagine there are more
[03:26]
_Dave_you -could- get the environment variable from an apache virtual host bucket ;)
whenever I do suexec, I'm pretty dang anal about what's in the environment
[03:27]
SvenDowideityup [03:27]
gac410The trick is coming up with a solution that works for everybody out-of-the-box with a mere mortal admin. [03:27]
SvenDowideitSvenDowideit wonders how well setting and the apache.conf ENV helps run the script frmo the cmdline [03:28]
_Dave_you can also do that in .htaccess if allowed [03:28]
SvenDowideitbut then, that's not somethign enough people use
whereas id configure were to create the script files, with settings inline
we should get a little speed, and alot of control and accuracy?
[03:28]
_Dave_doesn't INIT run like begin but everywhere? [03:29]
SvenDowideitSvenDowideit fails to grasp that statement, but then, i desperatly should have eaten an hour or more ago
time to go kill the neighbour's dogs
[03:30]
_Dave_O.O
are you THAT hungry??
http://perldoc.perl.org/perlmod.html#BEGIN,-UNITCHECK,-CHECK,-INIT-and-END
hmm you can also call FindBin::again();
[03:30]
SvenDowideitsomeone used FindBin, and i found a bug in it [03:33]
_Dave_oh lol
not surprising there
[03:33]
SvenDowideitas there are sadly too many odd corners of suexec, apahe, blah etc
all easy for the developer/admin type, no good for the user installer type
[03:34]
_Dave_well so you have a datapoint, a fellow geek friend of mine has refused to install foswiki because of the "." in @INC [03:34]
SvenDowideitlol [03:34]
_Dave_He read me the riot act and said I should know better lol [03:34]
SvenDowideithe could of course have made a trivial change to get it to work the way he prefers [03:35]
_Dave_well this guy is more anal than I am
his logic was "if THAT is there, I wonder what ELSE is there"
[03:35]
SvenDowideitor even better, been a useful member of the world, and submitted a better way to do it [03:35]
_Dave_hey, I don't get in the way of anality [03:35]
SvenDowideitanal, and a /user/ rather than a doer [03:35]
_Dave_sadly, I was forced to agree that he has a point, but I told him the developers were trying to change that [03:36]
SvenDowideitcourse, i have to laugh at that kind of ignorance, as i'll be he's not read every line of code in the OS's he runs [03:36]
_Dave_um...he has =P [03:36]
SvenDowideitunlikely
as there is rather s shedload of code both in kernel, libs and the rest of the toolset
[03:36]
_Dave_ok maybe not every line of code, but I've never found a line of code he didn't know about [03:37]
SvenDowideitso, snigger? [03:37]
_Dave_eh? lol [03:37]
SvenDowideitmeh - he's willing to use perl, but thinks he groks is? [03:38]
_Dave_I wouldn't take this personally or at all seriously other than as a datapoint contrasting the one where "foswiki has to install on any moronic web hosting company's obscure setup"
I'm one of those "there's two sides to everything" people :D
[03:39]
SvenDowideitthere's a million reasons not to use something - we're rather more interested in stepwise improving, otherwise we'd have to give up [03:39]
_Dave_true
the problem you are trying to solve has been around for some time
[03:39]
SvenDowideitand given there is only one tool that does what foswiki does, its not like we can jump ship [03:40]
_Dave_and is complicated by the plethora of methodologies of hooking perl to webservers
jump ship? lol...-stepwise- thinking here ... as you suggest :D
ya know speaking of this...my newest strategy is proxying to a pure perl webserver
because I'm sick of programs in between perl and the web browser
[03:40]
SvenDowideiti've been waiting for someone to finally see the light [03:41]
_Dave_just...don't use the POE one LOL
it works but does not scale in any way
[03:41]
SvenDowideitapplication servers are supposed to be behind a proxy, not directly attacked
i want to use the new & sexy - psgi
but i'm so overloaded its not funny
[03:41]
_Dave_I almost went Mojomojo because I like that strategy, but it was less mature than foswiki
what is psgi? and yes I'm overloaded too but I refuse to take that seriously
[03:42]
SvenDowideittotally different feature set
given that >60% of my waking time is looking after twins, i have little choice
[03:42]
_Dave_heh
that's the important stuff IMO
I don't have children of my own, I'm also in about 5 bands and they are full of people who resemble children LOL
[03:43]
SvenDowideitno idea :) given both of us are working half time from home, and have moved away from where our work was..
damn dogs just won't stop barking
[03:44]
_Dave_oo psgi [03:45]
pharveypharvey just made a topic called CritterQuery... a much more satisfying name than most of my other reports :) [03:45]
_Dave_interesting, but at first glance immature [03:45]
SvenDowideitpharvey, giggle
how's restplugin coming along?
commitable?
[03:46]
pharveyjust on my last catchup item now, then back to jqgrid + restplugin [03:47]
SvenDowideityay to both :) [03:47]
_Dave_if I move jquery to the latest version, how badly will that break the jquery table plugin? :D [03:47]
SvenDowideityou'll have to tell me :p [03:48]
_Dave_heh
hey I'll keep my eye on psgi, when it matures in a year or so I think that'd be great to use
[03:48]
SvenDowideiti think its already passed taht point
its a few years old
note the careless use of 'think'
[03:48]
_Dave_hee [03:49]
pharvey_Dave_, MichaelDaum has done a lot of work fixing things for 1.4.2 there's a task open somewhere (I think all items that we know about have been addressed) [03:50]
_Dave_nice [03:50]
pharveyItem:9739 [03:50]
foswikibothttp://foswiki.org/Tasks/Item9739 [03:50]
_Dave_because my %TABLE% macros wont turn off sort atm [03:51]
pharveyis this with JQTablePlugin? [03:51]
_Dave_yeah [03:51]
pharveyPatternSkin? [03:52]
_Dave_that item got pushed on my stack
I use pattern skin yes
and my own skin
SKIN = myskin, paternskin
*patternskin
[03:52]
pharvey"... latest jquery-1.4.x release is jquery-1.4.4. This release changed some semantics the way width() and height()..."
The jQuery guys have a more liberal understanding of 'patch release' than we do :)
1.4.0 -> 1.4.1 broke JSON parsing too
[03:53]
_Dave_oops
I've been playing too much with Dojo, have to swap the JQuery stuff back for this
[03:53]
..... (idle for 20mn)
***MTempest has joined #foswiki [04:15]
Lynnwood has quit IRC (Quit: Lynnwood) [04:22]
...... (idle for 29mn)
MartinCleaver has quit IRC (Quit: MartinCleaver) [04:51]
................. (idle for 1h20mn)
gac410 has left [06:11]
pharvey has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.14pre/20101201062444])
GithubBot has joined #foswiki
[06:16]
GithubBotfoswiki: master SvenDowideit * bf79a38 (2 files in 1 dirs): Item2052: improve the css-ability to showhide actions ...
foswiki: master SvenDowideit * 526fa28 (0 files in 0 dirs): Item2052: remove git repo - its all moved and strucured differently now ...
foswiki: master SvenDowideit * 6605fc1 (2 files in 2 dirs): Item2052: update versioning ...
foswiki: master commits d1412e6...6605fc1 - http://bit.ly/fp1rKl
[06:16]
***GithubBot has left [06:16]
MTempest has quit IRC (Quit: Leaving) [06:27]
favioflamingo has quit IRC (Ping timeout: 265 seconds) [06:33]
........... (idle for 51mn)
MTempest has joined #foswiki
MTempest has quit IRC (Changing host)
MTempest has joined #foswiki
[07:24]
vardit has joined #foswiki [07:30]
CDot has joined #foswiki
MichaelDaum has joined #foswiki
MichaelDaum has quit IRC (Changing host)
MichaelDaum has joined #foswiki
[07:37]
verboese has joined #foswiki [07:45]
MTempestCDot: I would have raised my concerns earlier, but I hadn't thought of them yet. [07:50]
CDotyes, your point is well made; I checked the doc, and nowhere does it say "revisions are identified by non-zero, positive integers. Older revisions will always be numerically smaller than younger versions, but it cannot be assumed that revisions always start at 1, nor that successive revision numbers are sequential"
so I'm just adding that
[07:53]
@() was the best I could think of for @*. If you have any genius ideas that don't involve either postfix nor nonary operators, then I'm all ears. [08:00]
MichaelDaumCDot, one update: so X@Y now means get rev Y of topic X. so you abandoned @ as REF ? [08:09]
CDotyes
CDot started looking at rev numbers, and has found many uses of "r1.94785"
thught we'd got rid of all of those :-(
[08:10]
MTempestIs @(*) workable? or @(all)? [08:11]
MichaelDaumwhy @
why not something speaking
a function name ... real words
[08:11]
CDot@ is mnemonic. "@" means "at" [08:12]
MichaelDaumthats what Larry says as well ... all the time [08:13]
CDotthe parser does not support functions, only operators. [08:13]
MichaelDaumothers think of perl as of an army of ants crossing the screen top down [08:13]
CDot@(*) is *not* workable
I *like* @ in this role. It reads very naturally.
[08:13]
MichaelDaumoh no functions [08:14]
CDotwell, a function *is* a unary operator, operating on an array, so it does really, but don't tell anyone. [08:14]
MichaelDaumin that case I'd always prefer functions above lots of operators [08:15]
CDotMichaelDaum: you often have preferences that are not aligned with anyone else. As do I. As do most people. [08:15]
MichaelDaumeven more for heavy lifting stuff
CDot, that's why I tell you
[08:15]
CDotCDot notes that this proposal was on the table for some weeks before I implemented it. Shame the comments only come after it's checked in; but hey, that's life. [08:16]
MichaelDaumyou are the only person modeling this language drip by drip. that way feedback by us comes rather late compared to your coding effords [08:17]
CDotCDot resolves to slow down his coding efforts
no worries, changing an operator name is not a big deal
just a bit of a PITA as all the unit tests have to change
[08:17]
MichaelDaumCDot, it is really hard to get the overall picture of all of the QUERY language given that it is split up in lots of feature proposals with lots of comments by lots of people with lots of different tastes and opinions. [08:18]
CDotAgreed. There is no formal spec; nor do I have time to develop one :-(
perhaps the time has come to start doing this using RFCs
[08:19]
LavrCDot - there are so many uncommitted feature proposal that we are drowning in them. It is not until someone commits to implement a feature that most of us wake up. This is why the 14-day clock starts when the proposal is committed and not when a brainstorm thing starts. [08:20]
MichaelDaumeither this works out in some brilliant way ... or it dies the death of thousand stiches
MichaelDaum is afraid of death of thousand feature requests
[08:20]
MTempestIt would be useful to identify the first revision and the last one, e.g. as @(first) and @(last) , which would give scope for @(first .. last). However, I don't like "first" and "last". Those words sound far too useful as field names. [08:22]
MichaelDaumbasically I can't really develop any well funded opinion whether @ or some rev() function is better ... because I LACK the overall picture [08:22]
CDotMichaelDaum: overall picture of what? How Foswiki works? How queries work? How queries map to SQL/XQuery/Mongo? How quantum states collapse? [08:24]
MichaelDaumMTempest, how about :first, :n-th, :last etc similar to jquery's selectors [08:24]
CDotCDot can't develop a well funded picture either, as he has done this work without any funding at all.
I tried not to narrow the syntax so far as to disallow things like :first and :last - these are good ideas, and could be easily integrated into the existing syntax
[08:24]
MTempestMichaelDaum: That works for me. [08:26]
MichaelDaumCDot, I mean the overal picture of the QUERY language. not women, good and the universe [08:26]
CDotMichaelDaum: where does System.QuerySearch fail? [08:26]
MichaelDaumit fails because it is docu of already *implemented* features ... not a discussion about possible alternative directions
once it is carved into docu stone ... nobody will bother ... most will say "oh well leave it like it is. dont touch"
even when people do have better ideas
[08:27]
CDotthat's true. However ideas are something we have never been short of. Implementations, on the other hand...... [08:29]
MichaelDaumCDot, sorry that I always have to give you this kind of level of a feedback ... but thats what it feels like being on the other side of the wire [08:29]
CDotmany idea sit there for a long time without attracting comment
it's only when someone implements something, that the comments come
and then the implementer is pilloried for not discussing the ideas!
[08:29]
Lavr(or put the name in the CommittedDeveloper field and a date) [08:30]
CDotwe discussed @() and the new query operators for a long time [08:30]
MichaelDaumI totally missed it [08:30]
LavrNot on a committed feature proposal. We notice the committed because most of us check the FeatureProposals topic regularly. But not all topic changes in Development web every day. [08:31]
CDotwhat can I say? Well, we have until 1.2/2.0 is released to change things, if the concensus demands it
CDot never considers his checkins to be "cast in stone". Collaborative development is all about helping eachother find better solutions.
[08:31]
MichaelDaumit is not that easy
reading the code and gettign the idea behind are two quite different things.
maybe @() is just fine...I simply do not know
[08:32]
CDotthat's why we discuss things in Development before implementing them [08:32]
MichaelDaumthat'S what worries me...maybe not your fault [08:33]
CDotwell, you have an implemntation to try now, and you can analyse and comment on it [08:33]
LavrI like the @. But MIchaelT had a good point about the need for the empty list result when you use a macro inside @(%MACRO%) where MACRO returns a list or interval. [08:34]
MTempestIt can take a long time for a proposal to get to the point where someone is ready to implement it, and the "current best idea" can vary considerably during discussion, investigation and brainstorming. I sometimes only spot things when someone says "I am willing to run with X". I hadn't looked at the code for @() (and still haven't). [08:34]
***Colas has quit IRC (Read error: Connection reset by peer) [08:35]
CDotMTempest: granted. And there is another issue that none of you have spotted. In order to implement @, I had to formalise the concept of "lists" in the query language, which has had knock-on effects on other operators. [08:35]
MichaelDaumme neither but I do see the direct value. e.g. reading the create date of the first revision....should be possible now isnt it? [08:35]
CDotThere is an open question in my mind as to whether +(1,2,3) should work, for example [08:35]
***ColasHome has quit IRC (Read error: No route to host) [08:35]
CDotd2n(1,2,3) is a better example [08:35]
***ColasHome has joined #foswiki [08:35]
CDotMichaelDaum: yes - *if* you can identify the first version
remember, version numbers *may not start at 1*
[08:36]
MichaelDaumI'd like to sort blog postings by create date using the @() operator [08:36]
CDotif you have a svn store, for example [08:36]
MichaelDaumis that possible? [08:36]
CDotI would be *delighted* if people experimented with @() *now*, rather than waiting for after release [08:37]
LavrAnyone implementing a store where revisions are not 1,2,3,4 etc should be shot. Any revision presented to the user must be 1,2,3 ... anything else is totally unacceptable in a real world. [08:37]
CDotLavr: yet SVN, git, manage quite well..... [08:37]
MTempestI side with Lavr on that one :) [08:37]
CDotthen you need to design a mapping. Cos right now, rev numbers (1) do not have to start at 1 (2) do not have to be sequential [08:38]
LavrSVN, git are not revisions shown to people. If someone implements a SVN storage you'd better make sure you map human revisions to the geek shit and hide the geek shit away from people. [08:38]
MTempestI think users could get used to SVN-style numbering, but I would hate to explain GIT revision numbers to my (CVS) users [08:38]
MichaelDaumit will be considerably hard to browse revisions if they aren't identified 1,2,3 [08:38]
LavrImagine a topic where first revision is 3455, and next is 4392 YUK! [08:38]
MTempestMTempest definitely prefers 1, 2, 3 to 47, 1030, 3192 [08:39]
CDotI was reasonably surprised that none of you commented when I raised this point before.
there is *no* internal mapping from "human" rev numbers (1,2,3) to "real" rev numbers (1234, 5678, 9101112)
[08:39]
MichaelDaumsay you need "get next rev number of object x"...for now this is currentRev +1 ... for correctness it should be nextRev(currentRev) ... whatever it takes to calculate nextRev() [08:40]
LavrI understood your comment as - it was not YET documented that revisions are 1,2,3,4 .... I could not imagine anyone even thinking that anything else would be acceptable. [08:40]
MTempestI did comment on the style of revision numbers two days ago on QueryTopicRevisions. [08:40]
CDotMichaelDaum: indeed; and the store supports that (getting the rev identifier of the next rev, and enumerating all revs) [08:41]
LavrAll this revision change stuff. Will it finally allow that I can QUERY for a field value based on a date? [08:41]
MichaelDaumCDot, it means: open up the repository and look it up ... could be quite expensive [08:41]
LavrIe give a date and get the topic revision before or after (or on) that date [08:41]
CDotLavr: sure. It has never been widely discussed because the RCS store uses 1,2,3
MichaelDaum: yes.
even determining how many revs you have is potentially expensive
[08:41]
MichaelDaumexpensive up to the point it is not feasible or usable [08:42]
CDotLavr: date -> rev no is actually one of the *easier* things to do
1,2,3 -> 1234,5678,9101112 is the hard bit
[08:42]
MichaelDaumwell, there are still tags supported by all systems
these are free form
[08:42]
CDot?
Lavr: you have been able to QUERY based on a date for a long time now
hwoever, you have *not* had access to old revs; which you do, now
[08:43]
MTempestheehee - tags applied to topics, CVS-style
:)
[08:43]
CDotthough wuith the RCS store, it is *expensive*
because to read an old rev, you have to ask RCS for t....
[08:44]
LavrCDot. What I meant was get to an old revision based on a date. Ie. I do not know the revision number but I have a date.
E.g. I want to query what a date field was in a topic on 01 Jun 2010.
[08:44]
CDotin the DBIStoreContrib, i am storing all revs flat, so the cost of getting an old rev is the same as getting the current rev [08:44]
MTempestMTempest thinks stores should support "get revision by date" [08:45]
CDotLavr: @()[date < d2n('1st Aprli 2010')]
MTempest: they do
[08:45]
MTempestExcellent [08:45]
***vardit1 has joined #foswiki [08:46]
CDot---++ ObjectMethod getRevisionAtTime( $topicObject, $time ) -> $rev [08:46]
LavrExample. I have a topic MyTopic with a field called Shipdate. What would be the syntax to show what the value of this field was on 01 Jun 2010? [08:46]
CDothowever in the query language there is no direct access to this operator
Lavr: you can do that, but only by jumping hoops
[08:46]
***vardit1 has quit IRC (Read error: Connection reset by peer) [08:47]
CDotyou *could* propose a "revat" operator [08:47]
LavrI was hoping your proposal would make this possible. [08:47]
MTempest@(revat(d2n('01 Jun 2010')) [08:48]
CDot(1) it wan't my proposal (2) it was discussed, but not committed [08:48]
***vardit has quit IRC (Ping timeout: 240 seconds) [08:48]
CDotCDot just implemented @(). it was proposed by Paul, IIRC [08:48]
MichaelDaumbbiab [08:49]
***MichaelDaum has quit IRC (Remote host closed the connection) [08:49]
LavrIn a business world being able to report formfields based on dates would be worth a lot. Senior managers always want reports with original dates, Last forecast (a week or a month ago) and current. This way they can see in reports of things are getting delayed. [08:49]
CDotLavr: I totally agree.
and a revat operator makes a huge amount of sense
[08:49]
MTempestNote - we could have a revat operator without nailing down the exact format of revision numbers. [08:50]
CDotindeed; in fact, it's a neat sidestep [08:50]
MTempestWe can define revat as returning "nonsense that is understood by @()" [08:51]
CDotif revs were always *displayed* by *date* and not *number*, the nature of the rev numbers would be moot [08:51]
LavrThe revat is also useful in itself. [08:51]
CDotindeed [08:51]
MTempestallrevs, firstrev, lastrev [08:51]
LavrI can create a list of dates in a topic. And these links point to the version of a topic at that date.
I can use this to list meeting minutes.
[08:52]
CDotin fact, to critics who say "1,2,3" I say "2010-01-30,2010-03-15,2010-07-02' ;-) [08:52]
***MichaelDaum has joined #foswiki
MichaelDaum has quit IRC (Changing host)
MichaelDaum has joined #foswiki
[08:52]
LavrTHis is another area where people create a copy of last weeks topic week after week. We have probably 2000 meeting weekly meeting topics now.
Instead people could just edit last weeks and reuse the same topic. The reason they don't now is that they want a list of the meeting minutes.
[08:52]
MTempestMTempest quotes CDot back at CDot: "If we're going to use dates (and I don't have a problem with that) then they should be specified according to ISO8601" :P [08:53]
CDotMTempest: 2010-01-30 *is* an iso8601 spec :-P [08:53]
LavrA list where you step the dates one week at a time and let it point to the rev at that time would solve that problem. [08:54]
CDotIIRC is08601 actually defined a syntax for date ranges [08:54]
MTempestMTempest goes to actually read the referenced spec :) [08:54]
CDotLavr: yeah; %QUERY begs the question why we don't have an n2d operator
it's mainly because IMHO GMTIME/SERVERTIME should handle it
you should be able to pass a epoch time to them.
[08:55]
***MartinRowe has joined #foswiki
AndreU has quit IRC (Ping timeout: 245 seconds)
[08:57]
CDotanyway, I look forward to proposals for :first and :last and revat - I think they are useful concepts.
and are all fully compatible with what's already there.
[08:57]
Babarok, I'll remove the use of FindBin, but... Gilmar used it first, and I never heard anything bad from him. I used to make it all by myself :) [08:58]
MTempestCDot: That still leaves the problem of differentiating between all revisions and no revisions. [08:59]
CDotMTempest: indeed; I look forward to your proposal for that as well
CDot is fresh out of ideas after his implementation marathon
[09:01]
MTempestUnderstood :) [09:01]
CDotCDot quite like MichaelDaum's idea of a special "class" of identifier, using a :prefix
I would even go so far as to say it should be :now
so, :now, :all, :first, :last all make (some sort of) sense
[09:02]
MTempestMe too. Hmmm. We should allow ThisTopic@(OtherTopic/WhatRevisions) - and the WhatRevisions field should be allowed to contain any of :now, :first, :all, :none [09:03]
***AndreU has joined #foswiki [09:03]
MTempestNot sure about :none
If we have :none, then @() can be an alias for @(:all)
Would we want to allow ThisTopic@(:OtherTopic/WhatRevisions) ? Note that the : is in the query, not in the field contents. If not, then one could always do ThisTopic@(:%QUERY{...}%)
[09:04]
***Babar sets mode: +oooo AndreU CDot ColasHome ktwilight_
Babar sets mode: +oo MichaelDaum MTempest
foswiki_irc9 has joined #foswiki
[09:06]
MichaelDaumError - unknown operattor :OtherTopic [09:07]
MTempestMichaelDaum: I agree.
Is there a way to say something like eval("%FOO%") || ':none' ? (please excuse my butchering of TML and perl)
I.e. to get :none if %FOO% returned an empty string
I could write a plugin to do that, but perhaps there is an existing way to do that? (SSP's CALC doesn't qualify)
[09:07]
CDotsome %FOO%s allow defaults
e.g. %URLFOO, %SEARCHFOO
[09:11]
MTempestDoes OR return the first operand, if the first operand is "true", otherwise the second operand? Or does OR always return a true/false value? [09:15]
***denisr has joined #foswiki [09:15]
MTempestMTempest is thinking @(%FOO% or :none) [09:16]
foswiki_irc9Hi, I'm not able to add a '*' to a URL string in the result of a formatted search. I want to construct an URL like <baseURL>?Name=foo*&Type=blah* where foo and blah are form variable results. [09:16]
MTempestOr even @( (%FOO%) or :none), if () makes a list out of what %FOO% returns.
foswiki_irc9 - that sounds like an encoding problem. Look at using the %ENCODE{...}% macro.
Something like %ENCODE{type="url" "foo*"}%
MTempest hasn't actually tried that
[09:17]
CDotMTempest: OR returns 1 or 0
all boolean ops return 1 or 0
[09:25]
***ArthurClemens has joined #foswiki [09:28]
MTempestSo I could do this: %QUERY{ "( ( field_a > 0 ) + ( field_b > 0 ) + ( field_c > 0 ) + ( field_d > 0) ) >= 2" }% to check if at least two of the four fields are greater than 0. [09:30]
***ArthurClemens has quit IRC (Ping timeout: 255 seconds) [09:32]
CDoterm... I guess so....
however mixing boolean ops and integer ops is not something I would even recommend
[09:38]
MTempestgiggle [09:41]
So... (1, 2) is a list. (1+2) is a number. Are there such things as a list-with-one-member, or a list-with-no-members? [09:46]
foswiki_irc9Mtempest - %ENCODE{}% doesn't seem to affect *, if I try %ENCODE{type="url" "http://hardware.slashdot.org/article.pl?sid=10/12/08/0222244*"}% the * at the end is unchanged.
I think the somehow the * is interpreted by formattedsearch, I've tried many variations of $formfield(Name)* like \*, '*', "*", \"*\," $nop* or placing the whole string betweeen \" \" but somehow the * allways seems to be interpreted as a special character.
[09:48]
***SvenDowideit_ has joined #foswiki [09:56]
MTempestAh. So you are searching for a topic that contains a literal "*" (plus some other stuff) ? [09:58]
***SvenDowideit has quit IRC (Ping timeout: 255 seconds) [09:58]
CDotfoswiki_irc9: since you are adding it to a URL, why not just use %42 ?
MTempest: the fact that $formfield() is immediately before it suggests it is not part of a search string
[09:59]
MTempestAh. I missed that. [10:00]
CDotthere is a list-with-no-members, () [10:00]
foswiki_irc9No the topic doesn't contain the *, but I need to have the * in result. [10:00]
CDotfoswiki_irc9: so use %42 [10:01]
MTempestCDot: Is a list-with-one-member any different from a single string-or-number? [10:02]
foswiki_irc9CDot:- That sounds good, I'm going to try that. [10:03]
CDotno...... ish. The interpretation of "an array" is really up to the operator that works on it
unary operators map {} the array
so -(1,2,3) => (-1, -2, -3)
[10:03]
***EugenMayer has joined #foswiki [10:05]
MTempestCan I have lists of lists? e.g. (1, 2), 3 [10:06]
CDotno
((1,2),3) == (1,2,3)
==(1,(2,3)
I did wonder about that limitation; but decided flattening the list was the lessed of the evils
[10:06]
MTempestMTempest wonders how to do a "default" operator, along the lines of default(foo, :none)... "(count(foo) > 0) ? foo : :none" ... ick [10:09]
CDotternary operators are.... tricky [10:10]
foswiki_irc9CDot- $percnt2A did the trick, thanks! [10:10]
CDot:-)
MTempest: TBH I'm tempted to throw out the infix parser and use a proper parser
I originally only used the simple stack parser because it was quick and easy
and a bison parser would be total overkill
something that sits in the middle would be nice.
(and is friendly to perl, of course)
[10:11]
***foswiki_irc9 has left [10:16]
pharvey has joined #foswiki [10:24]
pharveyCDot, hi. Did I get the syntax right on Development.QueryAcrossTopicRevisions? Maybe there's a bug [10:25]
CDot? [10:26]
pharveyReferring to examples @top
Maybe need [] instead of round brackets
Damn fiddly iPhone :)
[10:27]
CDotah, you are confused
the @ opertaor returns (one or more) meta objects for the specified rev
the / operator returns a single meta object for the latest rev of another topic
@()/ does..... what? I don't know
@()[condition] is probably what you wanted
[10:29]
***GithubBot has joined #foswiki [10:31]
GithubBotfoswiki: master CrawfordCurrie * e8adb8f (4 files in 1 dirs): Item10121: document the fact that revisions (1) may not start at 1 (2) need not be sequential (3) must be positive, non-zero integers. And if you didn't know that, then it's about time you did. ... - http://bit.ly/fFT7Pr [10:31]
***GithubBot has left [10:31]
CDotCDot notes that X/Y is probably synonymous with X@(:first)[|Y]
^:first^:latest
X@(:latest)[Y]
not sure about the precedence rules there. That needs to be checked.
[10:31]
***pharvey has quit IRC (Ping timeout: 265 seconds) [10:33]
foswiki_irc6hi there , i have install of strawberry foswiki and i am getting random 'the connection was reset' in my browser [10:36]
CDotsounds like a network problem/ [10:39]
***pharvey has joined #foswiki [10:46]
pharveyCDot: [] didn't help [10:47]
CDotpharvey: probably needs a closer look. Please annotate Tasks:10129 [10:47]
foswikibothttp://foswiki.org/Tasks/10129 [10:47]
CDotTasks:Item10129 [10:48]
foswikibothttp://foswiki.org/Tasks/Item10129 [10:48]
CDotCDot has to focus on wor$ today [10:48]
***pharvey has quit IRC (Ping timeout: 265 seconds)
AndreU has quit IRC (Quit: AndreU)
[10:51]
MTempestFWIW, http://en.wikipedia.org/wiki/Comparison_of_parser_generators lists a few that can output perl: http://search.cpan.org/~casiano/Parse-Eyapp/ http://www005.upp.so-net.ne.jp/kmori/kmyacc/ and Parse::Yapp
This thread could be useful: http://www.nntp.perl.org/group/perl.module-authors/2009/09/msg7844.html
[10:54]
Babaroutput perl? Why on Earth would one want that? [10:56]
MTempest(query) grammar -> parser-generator -> parser, which one can use in Foswiki
Feed the grammar for the %QUERY language into a parser generator, and get a parser one can use from Foswiki
[10:58]
CDotregexp::grammars looks interesting but the meta-language..... urgh [11:01]
***AndreU has joined #foswiki [11:02]
CDotI really ought to try and write up an EBNF for the query language, just for us geeks [11:02]
http://www.bbc.co.uk/news/technology-11928899 - sounds like the US govt needs a botnet. Perhaps they can borrow one off the Russians...... [11:09]
MichaelDaum^:latest^:last ... as in xpath and jquery
how do I query the previous rev, HEAD-1
:last-1?
[11:13]
MTempestHow about relativerev(:last, -1), which would work even if revision numbers are not sequential [11:21]
CDotwhat does :last mean? (I could read it as "the last before the current", "the current", or "the oldest") [11:21]
MTempestTo me, :last == :now, but I see your point. [11:22]
Babar@(HEAD..HEAD^)
or @(HEAD..HEAD~3) to get the last 4 revisions
[11:23]
CDot"head" is meaningless to anyone not steeping in revision control [11:24]
MTempestHEAD is geeky [11:24]
CDot"newest" is the message we want to get over. :fresh
CDot likes the idea of "use before" dates on topics
[11:25]
MTempest:current
:currant is like :raisin, but used for smaller topics. :grape is for fresh topics.
MTempest should get some more coffee
[11:25]
SvenDowideit_CDot I have a weird feeling about the listener meta constructor thing you commitde yesterady [11:27]
CDotSvenDowideit_: is it up your back? in your bowels? or manifesting as a sore toe? [11:28]
SvenDowideit_it feels to me like we'd be better off not having 2 kind of places that meta's can be instanciated
as what you've done is sidewards implement multi-store backend
[11:28]
CDotyeah, I know. which is why I left it as experimental
it's kinda like a back-door route I can use *right now*
[11:28]
SvenDowideit_ok - so you won't mind if in a month or so i totally move it [11:29]
CDotno [11:29]
SvenDowideit_ok, excellent
cos i too need that back door route in mongodb
but my plan was to re-jig it so that we don't do it that way :)
[11:29]
CDotit's really a "throw it in and see what explodes" job [11:29]
SvenDowideit_excellent - so it needs to be kicked and pushed and tossed away before we release from trunk [11:30]
CDotI'm sort of that opinion of the whole deep store listener concept [11:30]
SvenDowideit_unless is happens to be brilliant :) [11:30]
CDotbut it seems to be working out, somehow [11:30]
SvenDowideit_yup - its wrong and right at the same time
as it is essentially the replication concept that DB's use
[11:30]
CDotyeah; it just feels *too* deep [11:31]
SvenDowideit_imo, it should be merged into the logger stuff in a way too
but that is even more experimental (and happily not coded)
[11:31]
CDotI couldn't see how to delegate the ops to the Store superclass, because you need to know about the impl to know when to trigger the listeners [11:32]
SvenDowideit_in the sense that it dhould not be just listen
which is exactly what happens in database replication
for speed and efficiency (and for integrity) it is low level
[11:32]
CDotwell, my plan at the moment is to add revs to DBIStoreContrib (plus the new query ops) [11:32]
SvenDowideit_but that also is why its not a general api [11:32]
CDotand then look at caching prefs [11:32]
SvenDowideit_and then in the next month, something similar with mongo [11:33]
CDotas in, move in-text prefs into META: statements on the fly [11:33]
SvenDowideit_yup
fly/save/yup
[11:33]
CDotand then use DB queries
right
[11:33]
SvenDowideit_same plan, different timeline [11:33]
CDotI have no timeline [11:33]
SvenDowideit_and different flavour of ideas fror the api
you code faster than i
[11:33]
CDotthis is all pro bono, so has to be done when and if. [11:34]
SvenDowideit_and i keep having interruptions
took the girls to ikea on sunday
they ran around looking and playing with everything
and we just walked along behind them keeping them out of trouble/people
brilliant fun for all of us :)
[11:34]
CDotikea? fun? [11:35]
SvenDowideit_yup
the girls love interacting with stuff and people
and tbh, ikea is the most fun if you don't go to buy, but just to watch people
[11:35]
CDotit's Ok if it's not too crowded [11:36]
SvenDowideit_sunday was fantastic - as crowded as it can be
sunday, and a rainy day in a rainy week
[11:36]
CDotcould you still move? [11:36]
SvenDowideit_yup - the girls opened up all sorts of routes [11:36]
CDotah yes, you used the word "run" so not crowded at all [11:37]
SvenDowideit_just have to slip behind in their wake
people always make way for a running todler
[11:37]
CDotCDot has seen ikea so busy that even toddlers are trampled underfoot [11:37]
SvenDowideit_but yes, not as crowded as in a bigger city - brissy's only a few mill
that would be a sight :)
busy, not the trampled
[11:37]
CDotI'm told they have changed their routing policy too, with shotcuts and things [11:38]
SvenDowideit_yup
very much less mazelike
[11:38]
CDotued to be, you had to see *everything*, even if you only wanted an egg cup in the marketplace [11:38]
SvenDowideit_sadly
meh, time to go back to work, its only 8:40pm
[11:38]
MichaelDaum:last vs :latest ... given people understand :first, :last makes total sense as in "the opposite side of the spectrum" ... besides, it makes sense to look at similar languages ... like xpath, jquery, css dealing with sequences
either :first + :last two matching concepts ... or :earliest + :latest
[11:43]
SvenDowideit_or both? [11:44]
MichaelDaumthe first is more oriented towards addressing things being located somewhere on a sequence ... the second more wrt their temporal properties [11:45]
SvenDowideit_and so both make sense for different users at different times
they're just aliases/labels after all, and if they're in a limited context, the redundancy is unlikely to polute the global namespace
[11:46]
MichaelDaumyes, true. as long as both arent mixed
designing a language is all about choosing the right labels
[11:47]
SvenDowideit_:first and :latest feel ok to mix
:earilest and :last don't seem ambiguous either
[11:47]
MichaelDaumbetter don't mix them. for the same reason you wouldn't mix on+off with true+false [11:49]
SvenDowideit_mmmm, god, we need fo fix up that START and ENDSECTION vs BEGIN and whatever thing [11:49]
***GithubBot has joined #foswiki [12:01]
GithubBotfoswiki: master SvenDowideit * 79c239a (1 files in 1 dirs): Item2052: add a view css class if in view context - this really would be simpler if QUERY supported everything that can be IF'd or SEARCHed ... - http://bit.ly/gfmpT8 [12:01]
***GithubBot has left
pharvey has joined #foswiki
pharvey has quit IRC (Changing host)
pharvey has joined #foswiki
[12:01]
CDot has quit IRC (Quit: Leaving.) [12:09]
pharvey has left [12:18]
.... (idle for 15mn)
Babar sets mode: +ooo AndreU EugenMayer SvenDowideit_ [12:33]
Lynnwood has joined #foswiki
Lynnwood has quit IRC (Changing host)
Lynnwood has joined #foswiki
[12:44]
..... (idle for 21mn)
verboese has quit IRC (Ping timeout: 255 seconds) [13:05]
uebera|| has quit IRC (Remote host closed the connection)
Colas has joined #foswiki
[13:10]
.... (idle for 17mn)
m4nu has joined #foswiki
ArthurClemens has joined #foswiki
[13:28]
..... (idle for 20mn)
nutbar21Anyone here know a lot about the codebase for FormPlugin? [13:50]
***MartinCleaver has joined #foswiki [13:51]
nutbar21I'm having issues trying to solve this FormPlugin redirect problem... http://foswiki.org/Tasks/Item10133 - Maybe there's an easy workaround I'm missing? [13:58]
***Colas has quit IRC (Remote host closed the connection) [14:11]
ArthurClemensnutbar21: you'r Nick? [14:13]
***Colas has joined #foswiki [14:14]
nutbar21ArthurClemens: Yep. [14:14]
ArthurClemensI am author of the plugin. I have been struggling on that piece of code. [14:15]
nutbar21Me too :( [14:15]
ArthurClemensdo you have a suggestion for another code number instead of 307? [14:16]
nutbar21I've tried all the combinations I could think of
According to the HTTP 1.1 spec, all of the 300 codes should cause the browser to prompt the user, but as you see only Firefox does this (and we _don't_ want it to)
See "10.3 Redirection 3xx" at http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
[14:16]
ArthurClemensdid you change the number in the plugin code? [14:17]
nutbar21Yes
302 and 303 should work, but POST is lost and it tries to pass the form on to the save script via GET, which it doesn't like
[14:17]
***EugenMayer has quit IRC (Quit: Leaving.) [14:18]
ArthurClemenslooking at those protocols, would code 100 work? [14:20]
nutbar21I think I tried that. Let me try it again to verify. [14:20]
ArthurClemensnope [14:22]
nutbar21Hmm. It sent my browser, without prompting, to a plain text dump
It works if I disable form validation :|
[14:22]
ArthurClemensyes, the redirect is needed for form validation [14:24]
nutbar21That makes sense then.
Sounds like some JS would be necessary to fix this
[14:24]
ArthurClemensseems so
but it will a couple of weeks before I can dedicate time for this
[14:25]
nutbar21I found this earlier, maybe could give some ideas: http://stackoverflow.com/questions/46582/response-redirect-with-post-instead-of-get [14:26]
ArthurClemensthanks for the pointer. no, it won't be an easy fix
in your case I suggest to turn of validation
[14:29]
nutbar21Yeah, I might have to hack in some inline JS validation or something for the time being. [14:31]
***gac410 has joined #foswiki
terceiro has joined #foswiki
[14:35]
ArthurClemensI have updated the task with your info: http://foswiki.org/Tasks/Item10133 [14:36]
***favioflamingo has joined #foswiki [14:37]
nutbar21Looks good. Thanks for your input. [14:37]
MTempestI updated http://foswiki.org/Development/QueryAcrossTopicRevisions to reflect the discussion we had here. I hope I did it justice. [14:39]
..... (idle for 20mn)
***Bamie_ has joined #foswiki
Bamie_ has quit IRC (Changing host)
Bamie_ has joined #foswiki
GithubBot has joined #foswiki
[14:59]
GithubBotfoswiki: master GeorgeClark * 82968c2 (1 files in 1 dirs): Item10136: Add plugin module path to the diagnostics for admins ...
foswiki: master GeorgeClark * e05835f (1 files in 1 dirs): Item9715: Add attach listeners, and insert verb ...
foswiki: master commits 79c239a...e05835f - http://bit.ly/eK8wxi
[15:01]
***GithubBot has left
Bamieater has quit IRC (Ping timeout: 240 seconds)
Bamie_ is now known as Bamieater
[15:01]
foswiki_irc6hi there i have installed win foswiki 1.0.10 and i cant find any way of getting it to underline text am i missing something? [15:05]
gac410we currently don't have an underscore markup, so you need to use html. <u> text underlined </u>
You have to edit in raw mode, as the editor will protect the < >
Or with wysiwyg, you can enter the <u>text</u> and then highlight it all and select "Protect Forever" to protect the text.
Actually it appears you can just "Protect on Save" and the editor will handle it correctly going forward.
[15:07]
MTempestMTempest was about to say the same thing :) [15:12]
***wbniv has joined #foswiki [15:13]
ArthurClemensfoswiki_irc6: there is a reason: in html documents underlined text that is not a link is confusing [15:21]
***m4nu has quit IRC (Ping timeout: 265 seconds) [15:33]
foswiki_irc6thanks for your answers [15:41]
gac410foswiki_irc6: Note that there is an accepted proposal to add underlying syntax. Foswiki:Development/SupportLineThroughFormatTag But it's not slotted into a release yet. [15:43]
foswikibothttp://foswiki.org/Development/SupportLineThroughFormatTag [15:43]
***terceiro has quit IRC (Ping timeout: 240 seconds) [15:49]
CDot has joined #foswiki [15:54]
gac410CDot - I committed my changes to the Listeners API [15:55]
CDotgac410: cool. I still don;t understand autoattach
but I trust you have explored the alternatives
[15:56]
MTempestCDot: I added a summary of today's discussion to http://foswiki.org/Development/QueryAcrossTopicRevisions [15:57]
gac410We can remove / comment it out for now if you want. I have not done much exploring. The preferable alternative would be to deprecate autoattach and use a deferred topic update like Sven's UpdateAttachmentsPlugin
If we intend to mirror attachments over to a secondary store, auto-attachments that just show up and are not in the topic Metadata would be missed.
Because they are never actually touched by the attach api. They are "just there".
Even after a topic save, they are magically added to the metadata, but no attach listener would ever have been called.
[15:57]
***denisr has quit IRC (Quit: Parti) [15:59]
MichaelDaumMichaelDaum having problems to actually see what consensus was reached on http://foswiki.org/Development/SupportLineThroughFormatTag
note that XHTML is fading out. new kid on the block is HTML5
[16:00]
***CDot has quit IRC (Ping timeout: 255 seconds) [16:02]
MichaelDaum<del> is still valid html5
do we get new TML or not? this isn't clear reading the feature request
[16:04]
foswiki_irc6how can i enter strikethrough lines on text and secondly i notice the WYSIWIG editor will not put in double line spaces between paragraphs , so how would i enter those? [16:14]
***CDot has joined #foswiki [16:16]
CDot(16:00:26) CDot: gac410: the whole autoattach concept is so tied to a filesystem store, that it makes me very uncomfiortable
(16:00:44) CDot: MTempest: thanks. read it, good summary.
[16:17]
gac410Right. Just writing a small book in the task. I think by putting all autoattach as a separate verb, we avoid polluting the rest of the listeners with what is hopefully something we can deprecate completely. [16:18]
***favioflamingo has quit IRC (Ping timeout: 272 seconds) [16:18]
gac410I should raise a proposal to "DeprecateAutoAttach" and make UpdateAttachmentsPlugin part of core. [16:18]
***MTempest has quit IRC (Quit: Leaving) [16:18]
gac410Saved Item:9715 - Other option is to just drop autoattach now and say it is not supported by listeners. And write up the Deprecate proposal. [16:21]
foswikibothttp://foswiki.org/Tasks/Item9715 [16:21]
gac410Which TBH may be the better choice.
Need to make UpdateAttachmentsPlugin work on 1.1. though - it is broken. API to find the attachments was removed in the Store refactor.
[16:22]
LavrReverted the out of sequence docu nonsense. What the fuck have you been smoking CDot? [16:23]
CDotLavr: ?
CDot merely documents what the code does.
[16:24]
gac410gac410 steps slowly away from the laptop to avoid the backlash and flames ;-) [16:24]
CDotCDot gave up smoking many years ago. Perhaps that's the problem. [16:25]
LavrThe code does sequencial revs today. And unless an IDIOT change it it will continue to. [16:25]
CDotno. The RCS store happens to do sequential revs. There's nothing to dictate that anything else does them.
The SVN store, for example, never did sequential revs - which is one reason I never released it.
(aside from not seeing any real value in it)
[16:26]
foswiki_irc6i have put in autoattach and am looking at using this feature to load in hundreds of pdf files [16:27]
LavrThe SVN store is shit then. And anyone implementing a store that does not use sequencial numbers does not understand normal human beings. [16:28]
foswiki_irc6i hope it will work ok for me [16:28]
LavrAnd there is no purpose at all to document that someone moronic CAN implement a storage that does not present revs in sequence.
On the contrary we should make it clear that a storage MUST present revs in sequence.
[16:28]
CDotLavr: not at all. I can point to a number of stores that very successfully do not use sequential revs.
what you are concerned about is *presentation*, not *content*
[16:29]
LavrNot stores for end users. Do not come up with SVN and GIT here. They are not revision numbers of documents [16:29]
CDotfor example, if the skin presents revuision *dates* rather than revision *identifiers*
or if the UI allocates a sequential number to the stored revision.
[16:29]
gac410foswiki_irc6: autoattach is terribly inefficient, and often has to be disabled due to performance. A directory read is a very costly operation, and with autoattach gets done on every topic read. [16:30]
LavrWe should not encourage date instead of rev. Rev in sequence is the only way to ensure you have all revisions. [16:30]
***GithubBot has joined #foswiki [16:30]
GithubBotfoswiki: master KennethLavrsen * 4715426 (2 files in 1 dirs): Item10121: Revision scheme that allows jumping the revisions numbers ... - http://bit.ly/hJ8tVD [16:30]
***GithubBot has left [16:30]
CDotBut recovering an ordering from some stores is very hard, which is why we don't impose that constraint.
a _sequential numbering_, not an ordering, sorry
[16:30]
gac410foswiki_irc6: the alternative is a tool that runs the auto-attach in a shell script properly updating the topic metadata. UpdateAttachmentsPlugin - although currently broken on 1.1 [16:31]
LavrWe should impose it and that is why I did not revert. I changed the text so it is clear that revs are sequencial and start at 1.
And I will fight for that till I bleed.
[16:31]
CDotLavr: then you will have to provide the code to make it so, because that code does not currently exist [16:31]
gac410foswiki_irc6: we certainly can not revert Auto Attachments without providing a suitable alternative. [16:31]
CDotI am not defending non-sequential revs. I am merely documenting fact, rather than wishful thinking. [16:32]
LavrCDot. The RCS does this today. And anyone implementing other stores should ensure their implementation does that also. [16:32]
gac410foswiki_irc6: And by deprecate, it means that the option would still be there, but new features would not support it, and we would probably more strongly warn against it in configure.
foswiki_irc6: and of course any proposal has to be accepted after review.
[16:32]
LavrYou are documenting shitty UI. The code today that we have starts at 1 and goes 1,2,3,4. Fact. Anyone changing this = idiot. [16:34]
CDotLavr: perhaps. That doesn't make the code I am documenting work any differently. [16:34]
LavrYou are not documenting how the code works. [16:34]
CDotthere are other cases, not currently reaslied, where non sequential - even non-numeric - identifiers apply.
for example, changesets
[16:35]
LavrWe are dealing with revs of topics here. The revs people see in topics. And revs people use in SEARCH and QUERY. Do not mix anything else into the equation. And the docu we are talking about describes this. It is docu that end users read. And what was added makes no sense. I can already hear someone asking "How do I make a new revision that starts with 100?"
It is not a comment in the code. It is .txt files that end users read we are dealing with. And they need to be accurate to how the code works now. And in future a programmer that implements e.g. SVN numbers as revisions should be spanked.
If I edit a topic with revision 12 and save it I should not end up with 342.
And in a database store I assume the rev will be a field in a table and there - there will be no good reason to not go sequencial and start at 1 either. Why would you.
[16:36]
CDotLavr: the original reason for non-linear versions was the backend stores that people were talking about coupling in that do not support sequential (and even, in some cases, numeric) revision numbers.
The numbering scheme was deliberately decoupled from the core and pushed behind the store interface to permit this kind of thing
[16:43]
LavrProbably by someone that thought SVN numbers were a fine way to present revisions of a topic to people. [16:44]
CDotwhat a store exports to the core is up to the store. If you want to impose a 1++ constraint, then that needs to be documented in the store interface.
because currently, it permits any ordered sequence of integers.
[16:44]
LavrI want to impose a 1++ constraint to anything present to the end user. [16:44]
CDotnote that someone thought 1.1, 1.2 ... 1.120 was also fine [16:45]
***terceiro has joined #foswiki [16:45]
LavrI remember the Cairo days with 1.X and X++. Glad we got away from that [16:45]
CDotanyway, it is moot. You can document the constraint in the store interface if you wish. As you observe, it does not impact DB-based stores
and since that's the direction we are moving, I suspect non-sequential numbering can be treated as a red herring.
BTW the documentationw as a response to the criticism that nowhere did we say what a version number actually *was*
[16:46]
LavrIs does now with the checkin I did. [16:47]
CDotso we have uncovered something that we can calmly (and preferably politely) correct. [16:48]
ktwilight_ktwilight_ thumbs :) [16:48]
gac410Added FeatureRequest for 1.2 -Foswiki:Development/DeprecateAutoAttach [16:48]
foswikibothttp://foswiki.org/Development/DeprecateAutoAttach [16:48]
LavrYou can present it this positive way, yes. [16:48]
CDotgood. let's keep things positive.
we don't want these logs being picked up by wikileaks and splashed over the newspapers, after all.
I happened to note yesterday, when we were looking at the number of topics loaded in an average page view, that a number of standard plugins have $NO_PREFS_IN_TOPIC=0
CompareRevulsionsPlugin was one
[16:48]
LavrWe still have some left?
Let me check what they do.
[16:50]
CDotInterwiki was another; though that may be incurable
MichaelDaum observed that a "simple" topic view loads in excess of 60 topics. This is mainly access control related.
but anything we can do to reduce this burden is good.
[16:51]
MichaelDaumaccess control and webprefs [16:53]
Lavrgac410 - even though I am known as Mr. Anti-deprecate I am with you on the auto attachments thing. It is a misfeature used only by hackers that add files at file system level and such thing should happen when you add the file via scripts and not each time you view a topic.
CDot I saw the discussion about the many topic accesses.
[16:54]
gac410Yes, the UpdateAttachmentsPlugin works quite well on 1.0 - just needs to be reworked a bit for 1.1. [16:55]
***Colas has quit IRC (Ping timeout: 255 seconds) [16:55]
LavrI am not familiar with UpdateAttachmentsPlugin. [16:55]
gac410A rest script and shell script that can find and correctly attach files found in the Store. [16:56]
LavrAh. Sounds good. [16:56]
gac410There are in some cases very good reasons to bulk load up attachments. Maybe a needed feature would be to also allow it to read attachments from a source directory and move them in as well. [16:57]
LavrWe may even consider deprecation of auto attach without adding this as default. But it would need to be updated and better documented. And for sure also tested for non-A-Z attachment file names
ON the loading of the many topics I would think the creation of the web list in the left bar is one of the most expensive things we have.
If you have 200 webs you need to read and process 200 WebPreferences just to see if the web should be listed.
[16:59]
CDotcorrect
I (among others) have written various micro-caches just for that list
[17:00]
LavrThat could be cached in its own simple single file cache. List of webs incl access rights.
It is the fact that the list depends on the user that looks, that makes it more than just a flat file of webnames.
[17:01]
CDotI would prefer to address the whole problem of caching preferences, than attack it piecemeal
though the micro-cache approach does work adequately
[17:02]
LavrHow did you address the user dependency? [17:03]
CDotthe prefs DB has to be re-thought
currently the user dependency is intimate in the prefs DB :-(
in micro-cacheing, you simply cach the statically-evaluatable prefs, and complete the eval at runtime
i.e. bypass the prefs DB completely
[17:03]
LavrA weblist cache is easy to keep up to date when you create and move webs via the UI. If you do it at file level we would need some rebuild cache function. Could be a simple button in configure. [17:04]
CDotif the master plan comes off, we will be cacheing *all* prefs in the DB for fast eval
even the RCS store could be equipped with a cache
[17:05]
gac410SvenDowideit_: May I work on UpdateAttachmentsPlugin? (It says coordinate with author). [17:05]
***Colas has joined #foswiki
wbniv has quit IRC (Ping timeout: 255 seconds)
MartinRowe has quit IRC (Quit: Leaving.)
[17:18]
Babaroh for the record, I agree with Lavr that revisions should be presented to the user as consecutive
(and starting from 1)
[17:22]
***MichaelDaum has quit IRC (Remote host closed the connection) [17:22]
Babarit's easy to do in git, because anyway one would have to build the revisions by itself, so... :) [17:23]
CDotso do I. However that's not a requirement today, and I documented that fact. [17:23]
gac410by painting red concentric rings on himself ;-) [17:23]
CDotgac410: white, gold, blue, red and black. A toxophilists dream. [17:24]
Babaroh ok, so yes, it's not a requirement, but nobody in his right mind should do otherwise :) [17:24]
gac410:-) Laughing now that I looked up toxophilist [17:25]
BabarBabar thought bulls (how do one call that?) for bows had some blue in them [17:27]
***CDot has quit IRC (Ping timeout: 276 seconds) [17:33]
terceiro has quit IRC (Remote host closed the connection)
toffe82 has joined #foswiki
CDot has joined #foswiki
[17:40]
LavrI have an unusual requirement. Today we can use the parameter 'templatetopic' with the save script. This creates a new topic based on the topic defined by templatetopic. But if the topic you save already exists the 'templatetopic' is ignored. Then the content has to be in the 'text' parameter.
I am trying to find a way to save a topic that already exists and replace the whole content by the content in a template topic.
Anyone has hacked up something like this?
[17:48]
***ct529 has joined #foswiki
ct529 has quit IRC (Remote host closed the connection)
[17:52]
LavrToday I work around it manually by creating a new topic with a time stamp in the topic name and submit it to edit and not save. I then copy the whole content and cancel out. And then edit my baseline topic and replace the whole content by the paste buffer. I am trying to implement a more user friendly way to do this re-baseline action. [17:54]
CDotLavr: by the "whole content" do you mean the history as well?
or just the topic + form?
[17:55]
***terceiro has joined #foswiki [17:56]
CDotyou can (if you must) include the entire topic - including encoded meta-data - in the text passed to save
so if you passed the *entire text* (view?raw=all) it would achieve what you want.
[17:56]
LavrI need something slightly different actually. Let me explain the application. Maybe there is an even better solution.
I have a requirement spec which is built from a set of approx 250 individual requirement topics. The req spec is a nested search that builds a quite nice looking docu that people can print out and read like it was a word document.
But they want a base line document which is a static save of the nested searches.
[17:57]
CDotso use PublishPlugin [17:59]
LavrSo I do this by defining the requirement spec as a template topic with %STARTSECTION{type="expandvariables"}% around
I have not considered PublishPlugin for this. I will look at that. But the %STARTSECTION{type="expandvariables"}% method works fine for me. But my problem is that I have difficulty getting the normal users to use the method. It is a bit geeky to create a topic you never save and copy the content of the edit window.
PublishPlugin produces HTML. I need to produce revisions in TML. The idea of the baseline topic is that you can see the changes from rev to rev.
[17:59]
***wbniv has joined #foswiki [18:04]
LavrI probably end up having to write a BaselinePlugin to do what I want. [18:05]
CDotLavr: it sound more like a bastardised version of VarCachePlugin is what you want
but from your description, I would say a plugin is almost certainly required.
[18:06]
LavrThat is what I feared. But it should not be too difficult to make.
I could even just make a plugin that has a macro that take a topic and expands all macros but does not render anything and return the lot raw. I could feed that to the 'text' parameter of a html form submitting to save script.
[18:07]
It seems the Func::expandVariablesOnTopicCreation also expands "expandvariables" sections so such a plugin seems to be extremely easy to make. [18:18]
CDotyes. If you provide a postRenderingHandler you should get all you need (though you will have to tidy it up) [18:21]
***GithubBot has joined #foswiki [18:31]
GithubBotfoswiki: master KennethLavrsen * f147866 (1 files in 1 dirs): Item10091: Fixing docu of sections of type include so you understand that multiple sections are merged ... - http://bit.ly/egTU7o [18:31]
***GithubBot has left
CDot has quit IRC (Read error: Operation timed out)
[18:31]
GithubBot has joined #foswiki [18:46]
GithubBotfoswiki: Release01x01 KennethLavrsen * 2d1e9c3 (1 files in 1 dirs): Item10091: Fixing docu of sections of type include so you understand that multiple sections are merged ... - http://bit.ly/eXpIyn [18:46]
***GithubBot has left
uebera|| has joined #foswiki
[18:46]
MartinSeibert has joined #foswiki [18:58]
uebera|| has quit IRC (Excess Flood)
uebera|| has joined #foswiki
MartinSeibert has left
ct529 has joined #foswiki
uebera|| has quit IRC (Changing host)
uebera|| has joined #foswiki
uebera|| has quit IRC (Excess Flood)
uebera|| has joined #foswiki
uebera|| has quit IRC (Changing host)
uebera|| has joined #foswiki
[19:04]
...... (idle for 27mn)
Elron01 has joined #foswiki [19:36]
Elron01Good evening [19:37]
***bathroomninja has joined #foswiki [19:43]
Elron01I want to insert download-links for attached files. for 'unknown' files (like .doc) it works fine, but for known files, like .txt or .vbs the link open the content of the file in a new browserpage. I've looked at html-guides to find a parameter, but I don't found a hint. So how can I generate download-links for this files? [19:47]
bathroomninjaperhaps I'm just lost and stupid, but is there any install documentation? [19:47]
Elron01http://foswiki.org/System/InstallationGuide [19:48]
bathroomninjathank you! [19:49]
Elron01np [19:49]
gac410Elron01: Isn't that a function of the browser understanding the MimeType of the file being selected? [19:49]
Elron01and i there a way, to disable mimi in the link? [19:51]
***Elron01 has quit IRC (Quit: ChatZilla 0.9.86 [Firefox 3.6.12/20101026210630]) [19:52]
gac410Are your files served by apache from the /pub URL's? [19:52]
***Elron01 has joined #foswiki [19:52]
Elron01re [19:52]
gac410Are your files served by apache from the /pub URL's? [19:53]
Elron01for testing i uploaded a file on a test-topic and link that file with: [[%ATTACHURL%/ka.txt][ka.txt]] [19:54]
***bathroomninja has left [19:55]
Elron01but I dont know, where the file would be stored. mom, I take a look [19:55]
gac410If the attachment is served by Apache directly (your.host.com/pub/Web/Topic/Attachment) Then it's Apache that sets the mime type, through it's modules.
usually through mod_mime or mod_mime_magic
Though the browser will also look at the file signature and can choose to ignore the server provided type. ie. It may display a graphic if it recognizes a .gif file even if the mime type is incorrect.
[19:55]
***ct529 has quit IRC (Remote host closed the connection) [20:04]
Elron01(yes, the file would be saved in the pub-directory) but at other pages, there would be download links for files, that my browser maybe knows and could be combine with his mime-table. but on this pages, the 'download'-marked links open a the download-wizzard of the browser, indifferent, if this file is known by the mime-configuration of the browser. So I thought, there would be a way, to...
...force a link as a downloadlink, despite the mime-settings of the user-browser
[20:06]
gac410Unfortunately I don't know.
I found one reference, needs the "Content Disposition: " to be set to attachment. But that has to be done on the server.
[20:07]
Elron01np. i will search again for a awnser, and thanks for your effort. [20:11]
gac410http://www.thingy-ma-jig.co.uk/blog/06-08-2007/force-a-pdf-to-download
So it would be for all files that match, can't control it per link or page.
[20:11]
Elron01ok, thats ok, because for the wiki, mime isn't required. many thanks! [20:13]
***Colas has quit IRC (Remote host closed the connection)
Colas has joined #foswiki
[20:13]
ct529 has joined #foswiki
ct529 has quit IRC (Remote host closed the connection)
[20:23]
......... (idle for 43mn)
wbniv has quit IRC (Ping timeout: 245 seconds) [21:08]
........ (idle for 37mn)
wbniv has joined #foswiki [21:45]
.... (idle for 18mn)
pharvey has joined #foswiki
pharvey has quit IRC (Changing host)
pharvey has joined #foswiki
[22:03]
....... (idle for 31mn)
ArthurClemens has quit IRC (Quit: Leaving...)
MartinCleaver has quit IRC (Quit: MartinCleaver)
[22:34]
........... (idle for 50mn)
Demosthenes has quit IRC (Quit: leaving)
terceiro has quit IRC (Remote host closed the connection)
[23:27]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)