Release Meeting: 20 Mar 2017

(08:39:22 AM) MichaelDaum  entered the room.
(08:57:54 AM) vrurg entered the room.
(09:00:21 AM) gac410: Hi everyone.   
(09:01:23 AM) ***gac410 wonders if we need to consider a different schedule for release meetings.    To get better attendance.
(09:02:00 AM) gac410: We've not seen CDot or JulianLevens at a meeting for quite some time now  :(
(09:02:48 AM) gac410: Our agenda for today is same as the March 6th meeting -  which nobody showed up for.  
(09:02:51 AM) gac410: https://foswiki.org/Development/ReleaseMeeting02x02_20170306
(09:19:06 AM) MichaelDaum: hi all
(09:19:24 AM) gac410: Hi MichaelDaum
(09:19:28 AM) JulianLevens  entered the room.
(09:20:12 AM) gac410: Ah... and JulianLevens arrives too.    I was beginning to think we might not have a meeting this week either.
(09:20:34 AM) JulianLevens: Hi
(09:20:49 AM) ***gac410 has a time constraint.    Have to leave in 1-1/2 hours.   
(09:20:55 AM) gac410: Hi JulianLevens
(09:21:12 AM) MichaelDaum: I don't want to spent too much time on the RM either
(09:21:33 AM) gac410: We didn't have a meeting on the 6th,  so I'll use that for the agenda      https://foswiki.org/Development/ReleaseMeeting02x02_20170306
(09:21:53 AM) ***MichaelDaum has got the page open ... reading
(09:22:47 AM) gac410: The only new release blocker is the discovery of more unescaped { braces.   But Perl 5.25,  so not really urgent.
(09:24:39 AM) MichaelDaum: lets get started: 12511 is for me to comment
(09:24:57 AM) MichaelDaum: this is harder than expected
(09:25:25 AM) MichaelDaum: removing support for jquery.metadata might cause compatibility issues for our users
(09:25:36 AM) gac410: okay.  yeah  that's been hanging around for a long time.  Tasks that cross releases with commits into multiple releases are tough to figure out how to handle.
(09:25:51 AM) MichaelDaum: so I'd rather delay it for a 3.0 ... and still keep it in place while moving it to the deprecated plugins section
(09:26:45 AM) MichaelDaum: on Item14225 
(09:26:46 AM) gac410: well.    "Deprecation"  means that it's still there,  but recommended for not being used.   So can we "Deprecate" it in 2.2,  
(09:27:09 AM) MichaelDaum: yes okay
(09:27:36 AM) MichaelDaum: thats probably what I meant anyway ... hum
(09:29:00 AM) gac410: 14225 was my mess.    Just trying to decide if there was anything there worth keeping,  or just throw it away.    It was confusing that EXPORTEDPREFERENCES setting was inconsistent with what is exported.
(09:29:34 AM) gac410: Maybe it's just documentation to clarify what / when that setting is used,  
(09:30:14 AM) MichaelDaum: this is potentially an ash tray of things people amass without knowing what it was for
(09:31:04 AM) gac410: y  I was exporting preferences into the perl code,  then after the fact discovered the old setting, and was totally confused.
(09:31:04 AM) MichaelDaum: whenever a plugin or any other subsystem requires config settings being propagated from perl to javascript it should do so on its own
(09:34:15 AM) gac410: So for ex,   we needed COOKIEPATH and COOKIEREALM added to the exported preferences,    Wasn't I right to add it to the FOSWIKI.pm module where other prefs are exported?
(09:38:19 AM) vrurg left the room (quit: Ping timeout: 246 seconds).
(09:42:12 AM) gac410: While Michael's thinking  ...  We are just shy of a month left until our feature freeze for 2.2. ...  And there is a lot of work that appears to have not been started. :(
(09:42:27 AM) JulianLevens: Guilty your honour
(09:42:41 AM) gac410: JulianLevens:   You had two proposals ...  https://foswiki.org/Development/AddBacklinksToQuery and https://foswiki.org/Development/CleanUpFoswikiLocales   :D
(09:42:54 AM) JulianLevens: Actually, now I think about it. thats not quite true
(09:42:57 AM) MichaelDaum: gac410, yes I know. 
(09:43:23 AM) JulianLevens: Backlinks I was working on over a year ago, that's been fogotten :(
(09:43:26 AM) MichaelDaum: I am still on my backlog for clients
(09:43:40 AM) gac410: We are all guilty a bit.  ...I have several in my backlog that I've not started either. 
(09:43:42 AM) ***MichaelDaum was down by flu in feb
(09:44:13 AM) gac410: ugh...  flu not fun.   Hopefully you've recovered.  
(09:44:27 AM) JulianLevens: Locales I haven't touched for a while, but the code was complete IIRC, but the docs, the docs
(09:44:55 AM) gac410: For the locales,  don't we need to remove the "use Locale" from the begin blocks from all the modules? 
(09:45:46 AM) gac410: IIRC once we moved to using NFK in the sorting, it made the locales somewhat moot anyway.   
(09:46:15 AM) JulianLevens: No, arguably the proposal has changed, tidy up existing collation and deprecate use locale
(09:46:53 AM) JulianLevens: Vrurg has removed locale from 3.0 and I chose not to burden 2.x with possible subtle bugs
(09:47:01 AM) gac410: Ah ... okay
(09:49:12 AM) gac410: hm...  somehow I'm missing CDot's proposals from the backlog.   The Comments proposal for one.   It's awaiting revew & merge
(09:51:58 AM) JulianLevens: Finally divorced last Nov, motivation still comes and goes
(09:52:42 AM) gac410: https://foswiki.org/Development/SupportAllMacrosInTemplateTopics   Item13909   ready to merge.   Somehow omitted from the release plan :(
(09:56:30 AM) gac410: What we are missing from our development process is how we transition a "Ready to Merge" branch/item/feature   into master.
(09:57:02 AM) JulianLevens: y, I had wondered about that
(09:58:03 AM) gac410: When "Ready to merge"  the deveoper has decided their work is done.   So I suppose the rest of us have an obligation to look it over and then decide to merge at the release meeting.
(10:01:00 AM) JulianLevens: Well that's a step forward, I suspect something better is possible, I like the idea of creating a 2.1.x and 2.2.x branches etc, but without extra tools and developers I'm not sure that will work right now
(10:01:39 AM) gac410: "without extra tools" ?? 
(10:02:10 AM) JulianLevens: Hmm, not sure what I mean, except maybe CI to automate the whole thing
(10:03:40 AM) gac410: I think for now we merge features into master.   "feature freeze" stops merges of new features into master.  And then as we get close to beta status,  we branch to the 2.2.x branch
(10:05:53 AM) gac410: Then it's a joint decision on whether or not to stop maintaining 2.1 branch.  
(10:07:55 AM) gac410: Anyway JulianLevens ... backlinks is such a huge challenge.   Do you still want it on your plate?    Is there a simpler fix for now to get rid of the A-Za-z0-9 crap in the regexes?
(10:08:36 AM) vrurg entered the room.
(10:09:35 AM) JulianLevens: Leave it there, I'll have to report back closer to the FF date, it'll depend more on getting a burst of energy than anything else
(10:10:50 AM) JulianLevens: I cannot think of a worthwhile simpler fix
(10:11:56 AM) gac410: okay
(10:12:40 AM) vrurg: Hi everybody!
(10:12:51 AM) gac410: CDot has anoter one ready to merge.  https://foswiki.org/Development/ImproveSupportForComments
(10:12:58 AM) ***vrurg forgot about the meeting. 
(10:13:21 AM) gac410: Howdy vrurg
(10:15:07 AM) gac410: Actually for CDots ImproveSupportForComments,  he never set a committed date,  so propoasl is stuck in limbo.  https://foswiki.org/Tasks/Item13905 is ready to merge
(10:16:58 AM) gac410: vrurg,  have you settled on a minimum perl for the 3.0 code? 
(10:18:18 AM) ***gac410 wonders if in 2.2 we should announce end of support (deprecation) of Perl 5.8.8,   with minimum perl 5.x.x   in next major release.   5.18??
(10:18:57 AM) JulianLevens: I like that idea
(10:19:23 AM) vrurg_  entered the room.
(10:19:28 AM) vrurg left the room (quit: Ping timeout: 240 seconds).
(10:19:28 AM) vrurg_ is now known as vrurg
(10:19:42 AM) JulianLevens: There is a proposal out there related to the perl we should support
(10:20:17 AM) gac410: y,   I was looking at it,  and at the distrowatch perl versions for major distros.    (*#*@$   rhat 
(10:22:47 AM) gac410: RHEL is still 5.16.3 on their latest LTS   Which IMHO is too old.  We should be at least at 5.18,  
(10:25:13 AM) gac410: Centos is only 5.16.3 as well.  
(10:25:16 AM) JulianLevens: I'd like 5.18 as well
(10:25:41 AM) JulianLevens: But as we advertise and an 'Enterprise' wiki, I think 5.16.3 is the best we can do
(10:25:46 AM) gac410: For best unicode performance we really ought to be at 5.20   
(10:26:10 AM) gac410: vrurg,   iirc you might have required 5.18?
(10:26:19 AM) JulianLevens: So, we support 5.16.3+ and recommend 5.20+
(10:26:44 AM) vrurg: I have `use v5.14` in hte code.
(10:27:01 AM) vrurg: But would gladly raise it to 5.20. :)
(10:29:00 AM) gac410: I agree because of the laggard RHEL,  we could/should probably go to 5.16.3,   and make a statement in the release notes that the next Major release will require perl 5.16, but we strongly recommend 5.20 for best performance. ,.. or somethign like that.
(10:29:38 AM) gac410: Unfortunately distrowatch doesn't have a "what perl is supported across the distros" type of query. :(
(10:30:18 AM) JulianLevens: Ok, I need to go, thanks for the meeting
(10:30:23 AM) vrurg: gac410, it's unrealistic to keep track of every distro. Just ~10 of the top would be enough.
(10:30:29 AM) gac410: Thanks JulianLevens
(10:30:36 AM) gac410: vrurg:   I agree.   
(10:31:08 AM) gac410: afaik  rhat is the laggard.   SuSE is similar.
(10:31:30 AM) vrurg: And since the two majors are playing with 5.16 – then this is our base for now and there is no other choice.
(10:31:39 AM) gac410: ubuntu/debian/gentoo  are all in good shape
(10:32:03 AM) vrurg: I just installed latest suse two days ago. Didn't check it yet though.
(10:32:55 AM) gac410: SuSE is really confusing.   There's LEAP and Enterprise and some others.   And from what I could find,  they actually moved backwards in perl at one point.
(10:34:04 AM) vrurg: I've got LEAP 42.2. This one is 5.18
(10:34:58 AM) gac410: btw ... If you have not noticed,  CDot has Item14323  branch - upgraded to latest TinyMCE editor.  Working pretty well.   One glitch I've found is it scrambles if you move from natedit back to wysiwyg.
(10:34:58 AM) vrurg: But whatever. As long as rhel is pulling us back we cannot resist it.
(10:35:06 AM) gac410: y  I agree.
(10:35:48 AM) vrurg: I'm very much lack of time last months, track almost nothing.
(10:36:30 AM) vrurg: Would be happy if finish the specs code by the beginning of summer.
(10:36:49 AM) gac410: Turned out bigger than you expected?
(10:37:10 AM) vrurg: And 'finish' doesn' include the configure gui – this is up to somebody enthusiastic to deal with.
(10:37:31 AM) vrurg: And bigger, and less time than I expected.
(10:37:56 AM) vrurg: Actually, the legacy specs are being parsed into the new data format.
(10:38:35 AM) gac410: Okay.   One more question.    Backwards compatibility on extensions.    The Extensions installers,  now will "redact" any package marked as "NotCompatibleWith" ...   *should* we consider just dropping compatibiltiy with Foswiki 1.x on the core extensions and flag them as incompatible?
(10:39:07 AM) vrurg: I guess so. 
(10:39:16 AM) gac410: As we add features it gets harder to keep extensions working across releases.
(10:39:40 AM) vrurg: No doubt. 
(10:39:50 AM) JulianLevens left the room.
(10:41:08 AM) vrurg: But it's still to early to make any decisions. Not before we get more than just me involved with 3.0
(10:41:47 AM) vrurg: I'm inevitably biased in my opinions and decisions.
(10:42:37 AM) gac410: Well,  even for Foswiki 2.2,   %SCRIPTURL macro is not backwards compatible in all forms  :(     Added support for rest subject= verb=   and jsonrpc namespace= ...
(10:43:02 AM) vrurg: BTW, if I suddenly disappear – this is most likely my connection is gone. Have problems with my ISP.
(10:43:23 AM) gac410: no prob ...   I think we need to wrap up shortly.   
(10:43:50 AM) gac410: Unfortunately not a lot of progress on 2.2 yet.    I'm hoping we don't have to delay,  but the feature list is still pretty thin.
(10:45:10 AM) vrurg: If you remember, I have introduced a feature-grained compatibility control in one of my branches. This must significally ease control over extensions. Only those dependant on a removed feature will become obsoleted.
(10:46:13 AM) gac410: The challenge is how to prevent configure from listing them (noise) and installing them.   I've got the listing part fixed    but if someone runs the installer from the shell they'll still get installed.  :(
(10:46:51 AM) gac410: What brings this up is someone last week clobbered a 1.1.9 prod install by installing incompatible PatternSkin ... didn't notice the warning 
(10:47:01 AM) vrurg: There is no solution for the old codebase. 
(10:47:39 AM) vrurg: I'm disabling a loaded extension if it's not compatible. 
(10:47:43 AM) gac410: Well i've sort of fixed it.    Added a macro on f.o,   extracts the foswiki version from the Agent string reported by Foswiki::Net   and omits any extension 
(10:48:27 AM) gac410: the issue isn't the loading vrurg,  it's the other pieces.    The user installed PatternSkin which added incompatible templates.
(10:49:26 AM) vrurg: Ah, damn... Missed the point.
(10:50:07 AM) vrurg: So, there is no solution for this. Though the installed could and must perform compatibility checks.
(10:50:29 AM) vrurg: Ok, I have to go too. Thanks!
(10:50:31 AM) gac410: There is a PREINSTALL exit that can be used by the installer to block certain files.
(10:50:36 AM) gac410: okay   thanks vrurg.   
(10:50:40 AM) gac410: see you around...
(10:50:40 AM) vrurg: cu later.
(10:50:58 AM) vrurg: I'm usually on the chat anyway. Just watching.
(10:51:15 AM) gac410: anyway ...  Almost 2 hours in to meeting.   Let's adjourn.   Thanks everyone.