#foswiki 2014-01-14,Tue

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

WhoWhatWhen
***ChanServ sets mode: +o Lynnwood [03:49]
....... (idle for 30mn)
ChanServ sets mode: +o pharvey [04:19]
..................... (idle for 1h42mn)
gac410 has left [06:01]
................... (idle for 1h30mn)
ChanServ sets mode: +o CDot [07:31]
............. (idle for 1h0mn)
ChanServ sets mode: +o pharvey [08:31]
GuilainChello, new day, new hope
day 1 after 1.1.9 upgrade
imagegallery doesn't make thumbnails althougth http://foswiki.org/Tasks/Item9316 is done
editchapter work !
for imagegallery, seems to be me... my rewrite rules is not take into account
[08:45]
foswiki_irc0hello, anybody there ?
anybody who could answer a foswiki qustion?
[08:58]
***ColasHome has quit IRC (Read error: No route to host) [09:00]
jastdepends on the question ;) [09:00]
***ChanServ sets mode: +o MichaelDaum [09:03]
foswiki_irc0ok - i try it as short as possible: Have created a new Topic in the "Web Left Bar" - where I have to write the AccessControlString in - like the "Set ALLOWTOPICVIEW"-Statement [09:08]
jastyou want to protect the new topic? [09:20]
foswiki_irc0yes [09:22]
jastcreate the topic and add it in there as a bullet point (or, if you'd like to hide the statement, follow the "more topic actions" link in the topic, go to "edit topic preference settings", and add it there (three spaces, asterisk, Set ALLOWTOPICVIEW = ...) [09:22]
foswiki_irc0Thanks a lot!!! [09:25]
jastoh, and another space after the asterisk. basically a standard bullet point in TML syntax. :) [09:26]
............... (idle for 1h12mn)
JulianLevensCDot: QueryAlgorithms::query the comments suggest $webs passed but code does not, that itself is a minor point
However in the absence of a $web how should $inputTopicSet be interpreted?
[10:38]
OK, {web} is now passed via options. I suppose I have two key questions:
1: How should I interpret a mix of web names passed and web='xxx' in a query
2: Won't $inputTopicSet be empty in most cases, i.e. query across whole Foswiki database? If provided is it because the user gave a list of names?
[10:47]
***ChanServ sets mode: +o Lynnwood [11:00]
.......... (idle for 46mn)
JulianLevensCDot: As I've read more code it's slowing making sense although I do feel like a numpty at the mo [11:46]
CDotSven changed the semantics of the call. For Release01x01 passes in a topic set to search, but release 1.2 filters the search results post facto [11:47]
JulianLevensI figured something like that, thanks [11:48]
CDotbasically the input topic set has to be supported if you want 1.1 compatibility (DBIStoreContrib does that) [11:49]
JulianLevensI don't want that, or at least I have enough to think about [11:51]
..... (idle for 21mn)
CDotsure. You may be able to crib the code in DBIStoreContrib (not sure if the checked-in version has that, but the new version does) [12:12]
JulianLevensI've been targeting trunk from the beginning, so it'll need more than just that to work with 1.1, and I don't know the differences anyway
BTW: I'm convinced that a perl regex to POSIX regex converter is somewhat non-trivial. I can easily recognise the hoisted REs and convert them to something OK and then post filter with the original perl regex
[12:16]
CDotUnderstood. Targeting trunk has definite advantages, as the store interface is a lot cleaner.
on regexes, I have written a "personality module" for each supported DB (currently MySQL, Postgresql, SQL Server and SQLite) that maps Perl regexes to native DB expressions
[12:20]
JulianLevensWow [12:22]
CDotfor MySQL and SQLite the module simply uses pcre. For SQL server, I have an untested interface that I googled. [12:22]
JulianLevensWhat approach did you take then? [12:22]
CDotfor SQL server? [12:22]
JulianLevensNo, I meant in general and you've mostly answered that: use better target regex engine, i.e. insist MySQL uses PCRE engine? [12:24]
CDotyup [12:25]
JulianLevensPostgreSQL has TCL ARE which is pretty similar, are you passing that you or performing some simple conversions [12:26]
CDotthat's only if at least one SEARCH uses complex regexes. If ~ and ~= are not used, it will survive on ANSI SQL expressions.
simple conversions; example:
my ( $this, $lhs, $op, $rhs ) = @_;
$rhs =~ s/\\x([0-9a-f]{2})/_char("0x$1")/gei;
$rhs =~ s/\\x{([0-9a-f]+)}/_char("0x$1")/gei;
return $this->SUPER::regexp( $lhs, $op, $rhs ) if ( $op eq '~' );
# Postgresql supports full POSIX regexes. Just need to escape
# single quote.
$rhs =~ s/'/\\'/g;
return "$lhs ~ '$rhs'";
[12:26]
JulianLevensThanks I'm sure I'll crib that sometime. I really need to push my latest to GitHub and then take stock [12:28]
CDotit's only roughly tested at the moment, and I anticipate there will be constructs used in %SEARCH that can't be mapped. I'm just trying to do the best I can with what's available. [12:28]
JulianLevensThat's all you can do, plus some docs warning people what to avoid (assuming people want to change Stores from time to time)
Just checking the notes made here: http://foswiki.org/Development/RegularExpressions
That approach will cover the main DBMS, SQL Server is supposed to be pretty complete and powerful and therefore hopefully close to perl
There will be nits and obscure databases with poor regex support, but as you say it's the best we can do
[12:29]
CDotSQL server is *awful*
by default it barely supports the ANSI standard. But it has *loads* of M$ custom tweaks.
they've really gone to town on the lock-in.
[12:41]
JulianLevensThat's not there usual style
s/there/their/
[12:43]
CDotyeah, the "what to avoid" was part of the original motivation for the RegularExpressions page. Sad thing is there are so many variants e.g. between SQLServer2008 and SQLServer2011, there are major changes
not just in regex support, in other things too
and that's before you start looking at the best way to write SQL for different DBs
SQL has to be one of the worst standards it's ever been my misfortune to work with :-(
[12:44]
JulianLevensYep, when I originally embarked on this I planned to support many DBs, but after much research dropped that, at least for now [12:46]
CDotwell, it's been a focus of my work, so we can probably share
I haven't done much on SQL quality or data typing
[12:46]
JulianLevensMySQL and MariaDB only - although my SQL is pretty vanilla [12:46]
CDotCDot has found Postgresql significantly more forgiving that MySQL
than
[12:47]
jastMySQL likes to implicitly convert your data into random other types, despite any losses in precision
(obligatory reference to http://www.youtube.com/watch?v=1PoFIohBSM4)
in that sense, postgresql is significantly less forgiving than mysql, but in the good sense
[12:54]
....... (idle for 33mn)
CDotI was talking about regexes, but I guess there are plenty of other "features" waiting to leap out of the dark. [13:29]
......... (idle for 41mn)
***ChanServ sets mode: +o gac410 [14:10]
jastthe SQL-iverse is full of "features", yeah... [14:16]
............ (idle for 56mn)
***gregg4567 has left [15:12]
............................................................................. (idle for 6h20mn)
ChanServ sets mode: +o pharvey [21:32]
........... (idle for 52mn)
HaraldJoerg has left [22:24]

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