#foswiki 2015-09-19,Sat

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

WhoWhatWhen
***NewHorizons has left "See #NewHorizons for info" [00:05]
...................................... (idle for 3h8mn)
favioflamingohi guys, hope it is worth something to those working on a data backend. i started documenting my sql stuff here: https://github.com/favioflamingo/wiki [03:13]
................. (idle for 1h24mn)
***gac410 has left [04:37]
................. (idle for 1h24mn)
ChanServ sets mode: +o CDot [06:01]
.................................................................... (idle for 5h39mn)
JulianLevensgac410: PatternSkin, PlainFileStoreContrib and TopicUserMappingContrib are missing a trailing '\' for continuing multiple lines
as a result the table doesn't display properly
I think it's also stopping things like %$VERSION% being built properly
[11:40]
.............. (idle for 1h7mn)
***ChanServ sets mode: +o gac410 [12:50]
............... (idle for 1h11mn)
ChanServ sets mode: +o Lynnwood [14:01]
....... (idle for 33mn)
JulianLevensgac410: I think we should delete these from github: ChildTopicsTag, CreateTopicTag, EmptyTag, ImgTag, InclTag, MetasearchTag, TitleTag [14:34]
gac410JulianLevens: yeah they do indeed seem very obsolete, and completely outside of any acceptable naming. [14:47]
JulianLevens: Thanks for noticing the broken tables on the 3 extensions. None are actually released yet, so isn't a prod. problem.
I don't think it should be stopping VERSION though. ... that comes from the ExtensionForm now, and not from the table.
[14:53]
GithubBot[distro] gac410 pushed 1 new commit to master: http://git.io/vns9H
distro/master cbfbe4b George Clark: Item13504: Fix missing continuations in history table...
[14:55]
***GithubBot has left [14:55]
FoswikiBothttp://foswiki.org/Tasks/Item13504 [ Item13504: Documentation changes Foswiki 2.0.x / 2.1.0 ] [14:55]
.......... (idle for 47mn)
JulianLevensgac410: You said above '... that comes from the ExtensionForm now, and not from the table'
I've written code to scan the tables correctly and therefore fix up the form's in all plugins
The Q is what happens to the table - if an entry is in the form then can we lose it from the table?
[15:42]
......... (idle for 44mn)
gac410JulianLevens: It should be an either / or. If the value is in the form (preferred), then it should disappear from the table. Otherwise the regexes in the reports will double up the info. [16:28]
JulianLevensThat's what I thought
There's quite a bit of clean-up possible for example:
Add-on Author: 14 Author 3 Author(s): 122 Author: 105 Authors: 16 Contrib Author(s) 1 Contrib Author: 5
Those are some of the variants on Author
Wouldn't it be better to standardise?
[16:28]
gac410Copyright as well. Need to be sure it either uses (c) or © never the character [16:32]
JulianLevensOk, I'll try to do that by the end of the day [16:33]
gac410Y. I made the regexes more generous in buildContrib for finding Copyright and Author. Another practice I've found is addition of a "component" version which breaks the tools.
ie. BlahPlugin Version: | 1.2 Javacript Version: | 4.5 Which results in Version 1.2 4.5 in the tool comparisons.
we should only have the Version: string in the table once.
er... or in the form.
[16:33]
JulianLevensI have a much better sub to deal with this anyway [16:35]
gac410It is traditionally a wildcard because some use Plugin Version Extension Version, etc. [16:35]
JulianLevenshttp://pastebin.com/KbE0eW9x [16:36]
gac410Some of them should be generated. Dependencies : should come from %$DEPENDENCIES% substitution.
The tables or forms in the repo should use substitution for the stuff that BuildContrib will insert. ... Except for extensions that are manually released.
[16:37]
JulianLevensy, I've been getting familiar with BuildContrub [16:39]
gac410much of the data in the Extensions web is normally generated during build.
One thing it was doing that was very bad .. in the latest, was inserting defaults for Author / Copyright. etc. Those are always manually maintained.
License too.
[16:39]
JulianLevensI was planning to add that sub to BuildContrib instead of existing code and I recognised the %$VARIABLE stuff as build generated
What about Support and SupportUrl?
[16:40]
gac410Y. it's the target "Stage" iirc. [16:41]
JulianLevensy, Stage [16:41]
gac410Those are seldom anything other than the Support web. But with github, there is no reason that an extension author coulnd't use github for support requests, etc.
Nothing says an extensions must be supported by Foswiki.org
[16:42]
JulianLevensHowever, nobody actually does go elsewhere, I've scanned all Extensions even the dead ones on github [16:43]
gac410Y. we have not seen much 3rd party activity yet. [16:43]
JulianLevensI am progressing to some sort of Autobuild and Continuous Integration tool [16:43]
gac410cool
I think the project really needs to think about the unsupported chaff in Extensions. they make us look bad.
[16:44]
JulianLevensI'm not sure it would not be better to insist on foswiki github and f.o for built extensions
Or you support both yourself
I'm generating reports of the repos cross referenced with the published extensions on f.o to spot errors
[16:45]
gac410Historically we didn't require svn for extensions, but those are a pain when it comes to fixing. Same for "built" vs. manually uploaded.
iirc DavidPatterson's exts. were uploaded without buildcontrib. There have been some frustrated maintainers that just update the zipfile and update the attachment.
So the released ext. is out-of-sync with the source.
[16:46]
JulianLevensIn a way that's part of the point. Place you Extension in github and it will be built and published automatically
Plus I'll scan and report for known structural issues
[16:47]
gac410We need to be careful There are a lot of unreleased updates that made it into our "trunk" mess and had not been formally released.
svn just gathered commits, and then dev's lost interest or whatever and left svn not ready to release.
We really should insist that "master" always be in releasable status. ... we can do that for distro, but tough to enforce for extensions ... a branch/merge strategy.
[16:48]
JulianLevensy, I'd like to treat github as the master database and generate from there, but that database needs info such as Active, Dormant so we only publish the active stuff
I'll need to use branches with structured names to know what to build or not
[16:50]
gac410We also can't autobuild the "default" extensions. ... They have to follow the release process and sync up with the core release ... with limited exceptions for urgent fixes.
I suppose every repo has a "default" branch, which defaults to 'master'. And ideally that branch is always operational. So anyone using github to download the default gets operating code.
[16:51]
JulianLevensI've checked on github that they are always 'master' in practice [16:53]
gac410Right. The automatic upload ... all of github was built by script, set the default to Master. [16:54]
JulianLevensand it should stay that way :) [16:54]
gac410We have a cronjob on f.o, that detects new repos, and automatically adds the hooks, sets a default description, etc.
I have some local scripts that will run through all the repos and make changes
[16:55]
JulianLevensy, I've been looking
Anyway I'll get back to my immediate issue one of my github scripts does not always retrieve the latest commit-id of a repo
JulianLevens debugging ...
[16:55]
....... (idle for 34mn)
***ChanServ sets mode: +o CDot [17:32]
................................................. (idle for 4h2mn)
ChanServ sets mode: +o Lynnwood [21:34]
........................... (idle for 2h10mn)
GithubBot[distro] gac410 pushed 1 new commit to master: http://git.io/vnZVA
distro/master bca91a0 George Clark: Item13697: Item13728: Use File::Copy::Recursive::rmove()...
[23:44]
***GithubBot has left [23:44]
FoswikiBothttp://foswiki.org/Tasks/Item13697 [ Item13697: With iso-8859-1 store, it's possible to create unreachable topics and attachments. ] http://foswiki.org/Tasks/Item13728 [ Item13728: Rename fails across symlinked directories ] [23:44]
GithubBot[distro] gac410 pushed 1 new commit to master: http://git.io/vnZwn
distro/master 3d80f8f George Clark: Item13728: Use File::Copy::Recursive::rmove()...
[23:46]
***GithubBot has left [23:46]

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