User Tools

Site Tools


2015-03:day-2015-03-05

ΞX2Go Development Meeting 2/2015

Who is taking part?

  • [all]h1, Mike#1, Mike#2, Mihai

Topics

Research

  • In future meetings we will use this time slot for research presentation

Retrospective

    • [all]Which tasks/topics are still open?
    • [mihai][✓ ionic, 2015-03-05]Status git.x2go.org
      • [mihai][✓ ionic, 2015-03-05]Who's got access to git.x2go.org
      • [mihai][✓ ionic, 2015-03-05]What changes did you made to git.x2go.org
  • ΞX2Go Build Environment
    • [mihai][✓ ionic, 2015-03-05]Short Introduction about where to find information about "Package Building"
  • ΞX2Go Bugtracker
    • [mike#1][✓ sunweaver, 2015-03-05]Which bugs have been closed with the last update?
    • [mike#1]Which critical bugs hit us since the last update?

Planning

  • Which tasks should be taken care of?
  • VcXsrv Maintenance
    • Current hosting on SourceForge.net of our VcXsrv versions is terrible.
    • Upstream (1 developer) is releasing new major versions, but is unresponsive.
    • I (Mike#2) propose that I to host our version of VcXsrv the way nx-libs is being hosted: on both GitHub under ArcticaProject and on code.x2go.org.
    • The benefit of GitHub is that patches from the community are more likely, and they have a good issue tracker
  • Fedora 19 Jenkins builds

Raw Chat Log

(14:04:16) Heinz-M. Graesing: ...3, 2, 1...
(14:04:21) Heinz-M. Graesing: Welcome to our 2nc ΞX2Go Development Meeting 2015
(14:04:39) Heinz-M. Graesing: Please respect that this chatroom will be used as conference room for the next hour and make sure that this communication will not be disturbed by any questions until the meeting is over!
(14:04:53) Heinz-M. Graesing: For our participants: please make sure that a statement/post of a person won't be interrupted! There will be one active spokesman at once and he will tell when his post will be completed (EOS - End Of Statement).
(14:05:37) Heinz-M. Graesing: For short annotions - the whiteboard page is still active: http://wiki.x2go.org/doku.php/wiki:development:planning:whiteboard
(14:06:23) Mike#1: _o/
(14:06:26) Heinz-M. Graesing: to seperate all meetings from annother, we've introduces this page
(14:06:28) Heinz-M. Graesing: http://wiki.x2go.org/doku.php/wiki:development:planning:start
(14:06:46) Heinz-M. Graesing: so this will be our TOC today: http://wiki.x2go.org/doku.php/2015-03:day-2015-03-05
(14:07:08) Heinz-M. Graesing: if you want to have a look on the past meeting, it can be found here: http://wiki.x2go.org/doku.php/2015-02:day-2015-02-05
(14:08:05) Alex|2 [~kvirc@dslb-092-075-124-012.092.075.pools.vodafone-ip.de] hat den Raum betreten.
(14:08:12) Heinz-M. Graesing: to give everybody some time to assemble and to get arround, I'd like to invite mike#1 to give us some feedback about his research regarding nx, the romete desktop and the universe...
(14:08:17) Mihai Moldovan: Hi Alex!
(14:08:26) Alex|2: Hi All
(14:08:28) Heinz-M. Graesing: (and everything else...)
(14:08:35) Mike#2: Ohai!
(14:08:36) Heinz-M. Graesing: Hi Alex!
(14:08:40) Mike#1: h1Org: thanks!
(14:08:41) Mike#1: Hi all!
(14:08:47) HalfEatenPie hat den Raum verlassen (quit: Quit: Has been eaten D:).
(14:08:56) Alex|2: did I miss something? sorry for comming late.
(14:09:05) Mike#1: Within the last weeks mikedep333, Ionic and I have started investigating nx-libs.
(14:09:12) Stefan Baur: [again, I might seem unresponsive occasionally - and maybe even most of the time - it's because I'm on duty and have to take calls and answer e-mails from customers]
(14:09:15) Mihai Moldovan: just the introduction from h1 with all the links
(14:09:18) Heinz-M. Graesing: just o make sure: Alex -> this meetings toc can be dound here: http://wiki.x2go.org/doku.php/2015-03:day-2015-03-05
(14:09:35) Mike#1: http://code.x2go.org/gitweb?p=nx-libs.git;a=shortlog;h=refs/heads/3.6.x
(14:09:47) Mihai Moldovan: sunweaver: sorry for not reviewing PR4, I wanted to do it today
(14:09:49) Mike#2: and whiteboard: http://wiki.x2go.org/doku.php/wiki:development:planning:whiteboard
(14:09:53) Heinz-M. Graesing: Alex|2: no - but mike will beginn with some research results...
(14:10:04) Heinz-M. Graesing: so mike: this is your white space:
(14:10:08) Mike#1: Ionic, mikedep333 and I have decided that it is time to start working on a real fork of NX, not this patch pile we have been maintaining on top of vanilla NX.
(14:10:22) Mike#1: so, we branched off and the new branch is called 3.6.x
(14:10:33) Mike#1: see above URL to code.x2go.org
(14:10:41) Mike#1: what we have so far...
(14:10:50) Pinkamena_D [~paul@c-50-182-143-125.hsd1.nj.comcast.net] hat den Raum betreten.
(14:11:13) Mike#1: we took the master branch of nx-libs.git (3.5.0.x series) and applied all debian/patches as individual commits.
(14:11:14) HalfEatenPie [~HalfEaten@unaffiliated/halfeatenpie] hat den Raum betreten.
(14:11:31) Mike#1: then mikedep333 performed an awesome piece of work.\
(14:11:44) Mike#2: :)
(14:11:53) Mike#1: He went through all the CVE's in X.org from the time NoMachine forked until now.
(14:12:08) Stefan Baur: +1 from me 
(14:12:10) Mike#1: He checked if a CVE applies to NX and rebased patches from various origins.
(14:12:23) Mike#2: Which is December 2005, X.org 6.9.0
(14:12:27) Stefan Baur: mikedep333, I hereby declare you X2Go's "Deep C Diver"
(14:12:32) Mike#2: :)
(14:12:40) Mike#1: these patches have been committed natively on top of the 3.6.x branch and added as patch files to 3.5.0.x.
(14:12:56) Mike#1: http://code.x2go.org/gitweb?p=nx-libs.git;a=tree;f=debian/patches;h=43bee20ab3e3dd196eeb13ee67880025072c2520;hb=2b2a02f93f552a38de8f72a971fa3f3ff42b3298
(14:13:05) Mike#1: for this we had to renumber the patch file scheme.
(14:13:11) Mike#1: patches now have 4 digits.
(14:14:13) Mike#1: on the 3.6.x branch we will attempt at working as cleanly as possible (through pull requests and the pseudo-code review features on Github), so that individual commits can be cherry-picked later and copied to the ddebian/patches folder of the 3.5.0.x branch (aka master branch on git.x2go.org).
(14:14:24) Mike#1: master on git.x2go.org and 3.5.0.x should stay in sync.
(14:14:32) Mike#1: proposed rules:
(14:14:56) Mike#1: 3.5.0.x is maintained on git.x2go.org, commits there are reviewed within the X2Go team.
(14:15:34) Mike#1: the 3.6.x branch is maintained on github (https://github.com/ArcticaProject/nx-libs) and we use the code review features there to make commits as clean as possible.
(14:15:51) Mike#1: release goals for 3.6.x:
(14:16:08) Mike#1: get rid of bundled non-X11 libraries (Xpm, Mesa).
(14:16:25) Mike#1: get rid of un-adapted-to-NX X11 libraries.
(14:16:56) Mike#1: imake cleanup of all Imakefiles, remove build logic of everything that is not NX related.
(14:17:11) Mike#1: (e.g. Xdmx server, XprtServer, XFontServer).
(14:17:19) Mike#1: achieved so far:
(14:17:34) Mike#1: libNX_Xpm.so -> libXpm.so (system lib)
(14:17:53) Mike#1: quite a bit of the imake cleanup (on non-yet-merged branches)
(14:18:10) Mike#1: work in progress (on my side): more imake cleanup
(14:18:19) Mike#1: getting rid of libNX_Xfont.so.
(14:18:31) Mike#1: i.e. using the latest Xfont.so as shipped by Xorg.
(14:19:10) Mike#1: also achieved: nxagent used to have a hard-coded version string (3.5.0). This version string now gets set at build time to the actual 4 digit version we set.
(14:19:45) Mike#1: https://github.com/ArcticaProject/nx-libs/commit/203d97147623b97a48c0f6456cc1de9d8364db03
(14:19:53) Mike#1: other goals:
(14:20:05) Mike#1: update NX_Xrandr to proto version 1.4
(14:20:24) Mike#1: (NX_Xrandr is highly adapted to how NX works AFAICT, so we will have to keep that one.
(14:20:27) Mike#1: )
(14:21:10) Mihai Moldovan: may I ask one thing?
(14:21:25) Mike#1: also, from the Arctica project we plan to match nxcomp: currently, the nxproxy -S <-> nxproxy -C connection only works via IP sockets. We need that to work via unix domain socket files.
(14:21:31) Mike#1: Ionic: please.
(14:22:00) Mihai Moldovan: one of the most pressuing issues we face (especially with GNOME) is our outdated GLX 1.2 extension, shouldn't that be high priority?
(14:22:32) Mike#2: Ionic, Qt5 apps and the LXqt DE require Xrandr 1.4 http://bugs.x2go.org/cgi-bin/bugreport.cgi?bug=637
(14:22:35) Mike#1: as I said to mikedep333 yesterday, my goals grow with my personal learning code on this NX code.
(14:22:36) Mihai Moldovan: that may or may not be solved by updating MESA, but I'm not sure
(14:22:58) Mike#1: Xrandr and GLX are independent from each other concerning protoversions.
(14:23:32) Mihai Moldovan: yep
(14:23:33) Mike#1: my idea is: clean up first (NoMachine just patched their stuff in at disabled several build options).
(14:23:51) Mike#1: then move everything out that can be provided by Xorg.
(14:23:59) Mike#1: then improve stuff that has to stay...
(14:24:12) Mike#1: before we improve stuff first that will get removed 2 weeks later.
(14:24:47) Mike#1: all over approach should be: use as much stuff as possible from upstream.
(14:25:03) Mike#1: for the 3.6.x branch, I also look forward, not so much backward.
(14:25:10) Mihai Moldovan: yes, that sounds reasonable. I'm just asking because GLX is a reason some people get pissed with X2Go, as they can't use their ubuntu GNOME
(14:25:28) Mike#1: so with the 3.6.x I disrecommend from making it available for every old Linux distro on the planet.
(14:25:50) Mike#1: I suggest, developing along what is in latest Fedora, Debian sid and openSUSE Factory.
(14:26:18) Mike#1: Ionic: everyone is free starting to investigate on how GLX can be renewed.
(14:26:27) Mike#1: no blockage from my side, not at all.
(14:26:55) Mike#1: I am just listing what I can provide at the current time, and GLX is not on the list of top prios here (also speaking for the Arctica team here).
(14:27:10) Mihai Moldovan: sure, I just didn't know that
(14:27:16) Mike#1: it is on the list, but we need to be further in the cleanup process, I reckon.
(14:27:23) Mike#1: yeah.
(14:27:26) Mike#1: questions?
(14:27:39) Mike#2: Isn't Xpm an X11 library?
(14:27:39) Mike#2: http://cgit.freedesktop.org/xorg/lib/libXpm/
(14:27:44) Mike#1: yeah.
(14:27:51) steveeJ hat den Raum verlassen (quit: Ping timeout: 244 seconds).
(14:27:56) Mike#1: it is both, I guess.
(14:27:57) Mike#2: 2. When you talk a bout using the latest X.org libraries
(14:28:03) HalfEatenPie hat den Raum verlassen (quit: Quit: Has been eaten D:).
(14:28:09) Mike#2: don't you mean using the Linux distro's version?
(14:28:37) Mike#1: mikedep333: at first, yes, we work against the Linux distro
(14:28:42) Mike#1: 's versions.
(14:28:48) Mike#1: (because it is easy for testing).
(14:29:01) Mike#2: ok
(14:29:14) Mike#1: but there may be a moment where we realize that libX<something> needs a tiny patch to work properly with nxagent.
(14:29:27) Mike#2: 3. I noticed that newer versions of X.org server are built with newer versions of the C standard
(14:29:40) Mike#1: then we need to push that into X.org and I have Keith Packards workd (need to remind him) to have his support when it comes to that.
(14:29:44) Mike#2: this complicated backporting security patches to nx-libs
(14:29:47) Mike#1: (Keith Packard is Mr. X.org).
(14:30:04) Mike#2: which is C89 I think, like X.org 6.9.0 was
(14:30:15) Mike#2: any comments on this?
(14:30:51) Mike#1: Ok. Release goal 3.7.x (maybe, maybe not).
(14:31:02) Heinz-M. Graesing: This was a lot of input... wow!
(14:31:04) Mike#1: we have to go through the complete Xserver code in nx-libs
(14:31:15) Mike#1: https://github.com/ArcticaProject/nx-libs/tree/3.6.x/nx-X11/programs/Xserver
(14:31:44) Mike#1: ... and rebase every folder in there with what we find in X.org.
(14:31:46) Mike#2: 4. You forgot to mention some important cleanup. Like removing the xfree86 ddx except for 1 file we needed. This valuable cleanup will help us get included in Debian for example!
(14:32:02) Mike#1: until we can drop everything and off Xserver/hw/nxagent to the X.org team.
(14:32:46) Mike#1: mikedep333: as long as we ship Mesa in nx-X11/extras, I don't think we are ready for Debian, yet.
(14:32:53) Mike#1: same applies to shipping the libs.
(14:33:05) Mike#1: but, let me recall what you refer to.
(14:33:05) Mike#2: right, but I am pointing out that you did a very important job
(14:33:10) Mike#2: and that it is one of the things we need to do
(14:33:19) Mike#1: before all the above, I did a major removal of non-built files.
(14:33:20) Kjetil Fleten [~Thunderbi@78.156.106.62] hat den Raum betreten.
(14:33:35) Mike#1: see http://code.x2go.org/gitweb?p=nx-libs.git;a=blob;f=debian/CODE-REDUCTION_CLEANUP-UNUSED;h=13354a8d0d2b5cbe524b69e8eb9ffa4f4a9d4304;hb=2b2a02f93f552a38de8f72a971fa3f3ff42b3298
(14:33:54) Mike#1: and similar files under debian/CODE-*
(14:33:58) Mike#1: (3.5.0.x branch)
(14:34:12) Mike#1: code size shrank...
(14:34:51) Mike#1: 129MiB (NoMachine), 93MiB (after oriop took out the broom... _o/) to 41MiB now.
(14:35:06) Mike#1: source tarball size went down from 21MiB to 6MiB.
(14:35:32) Mike#1: basically EOS, questions?
(14:36:08) Mike#2: How many patches does TheQVD seem to be contributing?
(14:36:20) Heinz-M. Graesing: will there be changes of the communication with nxagent?
(14:36:23) Mike#1: they contributed two IIRC.
(14:36:35) Mike#1: Android patch for nxcomp library and nxproxy.
(14:36:37) Mike#2: And are any of the freeNX developers contributing?
(14:36:44) Mike#1: and the nxproxy/nxcomp version patch.
(14:37:13) Stefan Baur: mikedep333: There is no active FreeNX development taking place any more
(14:37:19) Mike#1: h1Org: you mean the comminication proto between nxproxy -S / nxproxy -C?
(14:37:22) Mike#1: no.
(14:37:28) Mike#1: not at this time.
(14:37:48) Stefan Baur: mikedep333: I'm subscribed to their mailing list, last message is from November 2014
(14:37:54) Mike#2: gotcha
(14:38:01) Mike#1: I am still eager on pertaining the full NX compat stuff (channels in the proto for file sharing, audio, etc.)\
(14:38:24) Mike#1: nothing heard from FreeNX.
(14:38:25) Heinz-M. Graesing: sunweaver: so in the beginning the handling won't change - which is interesting for all all projects, already using nx
(14:39:00) Mike#1: we have to cool contributors: Heinrich and Ulrich (I hope I got the spelling correct).
(14:39:10) Heinz-M. Graesing: sunweaver: Does the OpenNX developer knows about what is going on?
(14:39:15) Mike#1: Ulrich has been maintaining his on version of NX since NX 1.x.
(14:39:26) Mike#1: but he is not that much available due to lack of time.
(14:39:30) HalfEatenPie [~HalfEaten@unaffiliated/halfeatenpie] hat den Raum betreten.
(14:39:42) Mike#1: no idea of OpenNX. No contact has been made.
(14:40:18) Mike#1: Heinrich is very interested in armhf builds of NXAgent.
(14:40:26) Mike#1: EOS, other questions?
(14:40:48) Mike#1: I plan to blog about NX 3.6.x after our meeting.
(14:40:52) Mike#2: great
(14:41:05) Mike#2: one related topic I want to discuss is releasing nx-libs 3.5.0.29
(14:41:28) Mike#1: back to h1Org first, I'd say.
(14:41:32) Heinz-M. Graesing: It would be nice to get some information about Ulrich and Heinrich!
(14:41:33) Mike#2: ok
(14:42:11) Mike#1: both are C/C++ guys and very much at home with debugging, valgring and that sort of stuff.
(14:42:16) Heinz-M. Graesing: sunweaver: I don't want to interrupt further questions on this task - so if there are some: please ask!
(14:42:29) Mike#1: I know them from mailing lists mostly, so check the archive for their posts.
(14:42:48) Mike#1: Every post smells of high expertise, but both are limited in time.
(14:43:07) Heinz-M. Graesing: if there are no further questions at the moment, we should go on on our toc...
(14:43:15) Mike#1: Heinrich wants to contribute on armhf builds, so it would be good to "employ" him and invite him in.
(14:43:22) Mike#1: EOS.
(14:43:29) Heinz-M. Graesing: sry for interrupting!
(14:43:35) Mike#1: np.
(14:43:40) Pinkamena_D hat den Raum verlassen (quit: Ping timeout: 252 seconds).
(14:45:13) Heinz-M. Graesing: It was a hounour to take part on the packages building last week - thank you sunweaver and Ionic. To make sure, that everybody get some information about that, would it be possible to point them to some pages and ressourcesß
(14:45:25) Heinz-M. Graesing: ?
(14:45:31) Mike#1: Ionic: ?
(14:45:39) Mihai Moldovan: yep
(14:46:04) Mihai Moldovan: http://wiki.x2go.org/doku.php/wiki:development:release-howto < this is the general release howto, which still needs a little bit of love regarding branches and debian/changelog cleanups
(14:46:38) Mihai Moldovan: mikedep333 has done a great job at setting up a similar document for windows builds: http://wiki.x2go.org/doku.php/wiki:development:release-howto-mswin
(14:46:51) Mike#2: :)
(14:46:57) Mihai Moldovan: both are linked on http://wiki.x2go.org/doku.php/wiki:development:start
(14:46:58) Mike#1: :)
(14:47:42) Mihai Moldovan: that's basically all on the information front
(14:47:55) Mihai Moldovan: what I'm still a bit unclear on is how often releases are to be done
(14:48:28) Mihai Moldovan: like, for x2goserver, I'd like to release yet another version, although the last release is only one or two weeks old
(14:48:48) Heinz-M. Graesing: Ionic: there will bne to trigger for taht: serve bugs and new features - I think the question is open to the developers...
(14:49:00) Mihai Moldovan: in the meantime, though, I fixed an RPM bug which affects users that install x2goserver for the first time (session db creation)
(14:49:20) Mihai Moldovan: that should finally be working automatically now
(14:50:12) Mihai Moldovan: mikedep333 hit the bug as soon as installing the new x2goserver version released x days ago in a clean system
(14:50:38) Mihai Moldovan: as this bug is making x2goclient crash, that's a very bad starting point for new users
(14:50:39) Mike#2: actually, I did an upgrade
(14:50:45) Mike#2: from 4.0.1.x to 4.1.0.0
(14:50:46) Mike#2: IIRC
(14:51:14) Mike#1: Ionic: did you fix the db creation in x2goserver.spec on master or on release/4.0.1.x.
(14:51:17) Mike#1: ?
(14:51:25) Mihai Moldovan: sunweaver: on master and on release/4.0.1.x, yes
(14:51:28) Mike#1: good.
(14:51:38) Mike#1: any non-packaging changes on release/4.0.1.x?
(14:51:47) Mihai Moldovan: the same goes for the mising (now implemented) /etc/x2go/applications symlink
(14:52:02) Mihai Moldovan: nope
(14:52:28) Mike#1: may suggest something?
(14:52:35) Mike#1: may I...
(14:52:56) Mihai Moldovan: sure
(14:53:12) Mike#1: don"t bump a new upstream release, simply increment the Release: field in nx-libs.spec and update the build-main branch.
(14:53:40) Mihai Moldovan: ah, ok
(14:54:04) Mike#1: then you have your fixes out there, but don't have to bother with a release.
(14:54:20) Mihai Moldovan: hehe, x2goserver.spec, not nx-libs.spec
(14:54:27) Mike#1: ahja, right.
(14:54:39) ***Mike#1 has been digging to much in the NX realm lately.
(14:54:45) Mihai Moldovan: yeah, that's reasonable
(14:54:57) Mihai Moldovan: so I'll increment x2go1 to x2go2
(14:55:08) Mike#1: yeah.
(14:55:30) Mihai Moldovan: or maybe even to 0.1x2go2
(14:55:31) Mike#1: do it in debian/changelog and in x2goserver.spec, so those stay in sync (with no gain on the .deb package side).
(14:55:35) Mihai Moldovan: err, 0.1x2go1
(14:55:47) Mike#1: -0x2go2
(14:56:22) Mihai Moldovan: why are you using 0.0x2go anyway? x2go to mark our upstream package, but 0.0?
(14:56:23) Mike#1: the 0.1 1.1 stuff in Fedora/RPM alludes to EPEL-X backports, in Debian it alludes to non-maintainer uploads.
(14:56:31) Mike#1: don"t confuse people there.
(14:56:34) Mihai Moldovan: ok
(14:57:02) Mike#1: it was some EPEL-related versioning style stuff I can hardly remember at the moment...
(14:57:35) Mihai Moldovan: well, good enough, I was confused myself
(14:57:56) Mihai Moldovan: so that's that, if there aren't any more questions on x2goserver?
(14:58:23) Mike#1: the question about how often release should be made is still open.
(14:58:39) Mike#1: it depends on feature / fixes.
(14:58:43) Heinz-M. Graesing: sunweaver: and I think there are two ansers
(14:58:46) Heinz-M. Graesing: answers
(14:59:02) Mike#1: ?
(14:59:04) Heinz-M. Graesing: features and bugfixes
(14:59:07) bigfoot-: hey, is there going to be a nxagent/x2goagent release soon? :)
(14:59:37) Mike#1: bigfoot-: shhhh (meeting)...
(14:59:38) Heinz-M. Graesing: to refer towards fixes: sessionsharing -> !!!
(14:59:52) Mihai Moldovan: bigfoot-: thanks, we'll keep track of that question (I forget about it)
(15:00:06) Heinz-M. Graesing: maybe we sould give a very short information about what is fixes now... (after the last release)
(15:00:11) Mike#1: h1Org: session sharing is fixed again.
(15:00:28) Heinz-M. Graesing: sunweaver: YES!!!
(15:00:40) Mike#1: http://permalink.gmane.org/gmane.linux.terminal-server.x2go.announce/161
(15:00:58) Mike#1: the info can be found in the release announcement, no need to repeat here, I guess.
(15:01:55) Mike#1: may I suggest something....
(15:02:00) Mihai Moldovan: go ahead?
(15:02:11) Mike#1: the plan is to meet every two weeks here, right?
(15:02:12) Heinz-M. Graesing: sunweaver: it is a answer on "when should we release" ...
(15:02:22) Mike#1: yse.
(15:02:42) Mike#1: the different components in X2Go are maintained by different people.
(15:02:52) Mike#1: and also that changes from time to time.
(15:02:54) Heinz-M. Graesing: sunweaver: I've already made a "doodle" on the planning page: http://wiki.x2go.org/doku.php/wiki:development:planning:start
(15:03:42) Mike#1: so, basically, if someone feels some X2Go component should be released (because of feature add-ons, minor fixes), bring it up here and Mihai/h1Org take care the weekend (or so) after.
(15:04:14) Mike#1: if any of us fixes a severe bug, a quick-release will be communicated via x2go-dev.
(15:04:15) Heinz-M. Graesing: +1
(15:04:39) Pinkamena_D [~paul@c-69-248-138-174.hsd1.nj.comcast.net] hat den Raum betreten.
(15:04:52) Heinz-M. Graesing: so the release can be discusses!
(15:05:03) Mike#1: nx-libs -> mikedep333.
(15:05:05) Mihai Moldovan: I'm just careful to re-release stuff only one week after the last release
(15:05:19) Mike#1: you can't release too often...
(15:05:24) Mihai Moldovan: exactly
(15:05:43) Mike#2: it's my turn to speak?
(15:05:57) Heinz-M. Graesing: about VcXsrv?
(15:05:59) Mike#1: you have done most of the work and testing on the 3.5.0.x branch, I guess, yes.
(15:06:03) Mihai Moldovan: well, we probably need a nx-libs release
(15:06:06) Mike#2: about nx-libs 3.5.0.29
(15:06:07) Mihai Moldovan: rather soonish
(15:06:12) Mike#1: no, about nx-libs 3.5.0.29 release.
(15:06:25) Heinz-M. Graesing: ok mikedep333 go for it!
(15:06:29) Mihai Moldovan: x2goclient and x2goserver are done, nx-libs is the one big component missing
(15:06:36) Mike#1: mikedep333: please report on your work and status.
(15:06:42) Mike#1: and test results.
(15:06:50) Mike#2: All the security patches were backported to 3.5.0.29, but we have not released 3.5.0.29 yet.
(15:07:03) Mike#2: I tested it on a few of my machines with my usual workflow. There were no regressions.
(15:07:04) Mike#1: any regressions so far?
(15:07:12) dgandhi1 hat den Raum verlassen (quit: Quit: Leaving.).
(15:07:27) Mike#1: Alex|2: do you feel like reviewing the patches?
(15:07:32) Mike#2: since users are lacking security updates, I believe we should release ASAP
(15:07:36) Mike#2: EOS
(15:08:00) Mike#1: Alex|2: ?
(15:08:06) Alex: sunweaver: you mean x2go patches?
(15:08:27) Mihai Moldovan: nx-libs patches
(15:08:36) Mike#1: http://code.x2go.org/gitweb?p=nx-libs.git;a=tree;f=debian/patches;h=43bee20ab3e3dd196eeb13ee67880025072c2520;hb=2b2a02f93f552a38de8f72a971fa3f3ff42b3298
(15:08:43) Mihai Moldovan: the CVE fixes from 3.5.0.28 to 3.5.0.29
(15:08:44) Mike#1: everything greater 1000_...
(15:08:56) Mihai Moldovan: (but I've basically done a review and fixed errors I could find)
(15:09:08) Mike#1: changelog is here: http://code.x2go.org/gitweb?p=nx-libs.git;a=blob;f=debian/changelog;h=c9dd06dd62738d9a0e978cfee7191723f96cee3a;hb=2b2a02f93f552a38de8f72a971fa3f3ff42b3298#l131
(15:09:09) Mike#2: Ionic, thank you :)
(15:10:13) Mike#1: Ionic: right, you did.
(15:10:37) Mike#2: also
(15:10:59) Mike#1: so, release, then!?!
(15:11:14) Alex: Did some one test it?
(15:11:25) Mike#2: I was particularly concerned about the glx code in nx-X11/programs/Xserver/ having regressions, but we have terrible glx support anyway and I tested that code with a few 3d apps. There were not regressions.
(15:11:30) Alex: are they available as nightly builds?
(15:11:34) Mike#1: I run the latest builds always on various sites.
(15:11:35) Mike#2: Alex|2, yes
(15:11:36) Mihai Moldovan: but if Alex wants, he can surely go through them again, do a sanity check and look for overlooked errors
(15:11:40) Mike#1: No observations of regressions, so far.
(15:11:50) Mihai Moldovan: Alex|2: we currently use 3.5.0.29 nightly builds
(15:11:53) Mike#2: Alex|2, I tested for regressions.
(15:12:20) Heinz-M. Graesing: Alex|2: the question about our testing is a very good question - and we really should take care of it!
(15:13:19) Alex: I think it's Ok, but I'll look on it once again, ok?
(15:13:29) Mihai Moldovan: sure
(15:13:33) Alex: So, I think relase beginning next week?
(15:13:38) Mike#2: The software industry in general is heavily focused on automated testing as part of devops/continuous delivery. Is that what you are talking about?
(15:13:43) Mihai Moldovan: release as soon as we hear from you
(15:14:13) Heinz-M. Graesing: mikedep333: yes - but it will need some time, to get it running on our project!
(15:14:19) Mike#2: yup
h1Org h1Org1 
(15:15:29) Stefan Baur: h1Org +1 on the automated testing. Suggest putting it on the Agenda for an April dev meeting or the Gathering, or both.
(15:15:29) Mike#1: h1Org: in Arctica, we will weave in unit testing and performance/duration tests from the beginning.
h1Org h1Org1 
(15:15:47) Stefan Baur: h1Org April, because I want to get LIT Augsburg done first
(15:17:09) Mike#1: ok, result: Alex|2 does another review, pings us via x2go-dev and Ionic will do the nx-libs release.
(15:17:20) Mike#1: beginning of next week should be the time for the release, then.
(15:17:31) Mihai Moldovan: maybe with mikedep333
(15:17:50) Heinz-M. Graesing: MyNameIsRetro: this will take much more longer and should be well planned - because we'll need some reports, visualation and monitoring too
(15:17:52) Mike#1: I will join in, because there are slight difference to the other X2Go components.
(15:18:09) Mihai Moldovan: hum, okay... that's more complicated then
(15:18:33) Mike#1: I also think that automated testing needs some real attention. we all feel like we need to get more settled in our new roles.
h1Org h1Org1 
(15:18:47) Alex: How about reducing the amount of X2Go Components, throw away things that not used any more, or have no future?
(15:19:01) Alex: kde3 admin tools, x2goplugin?
(15:19:05) Mihai Moldovan: Alex|2: most components are in active use
(15:19:16) Mihai Moldovan: Alex|2: x2goplugin has had a mailing list post recently
(15:19:17) Mike#1: -> agenda for next meeting, please.
(15:19:18) Stefan Baur: h1Org I'm just saying, important topic, let's not forget about it, but also let's not open too many issues at once, that's why I suggest the April date
(15:19:25) Mihai Moldovan: and a patch
(15:19:46) Heinz-M. Graesing: MyNameIsRetro: yes!
(15:19:46) Mihai Moldovan: (which I still need to adapt/merge)
(15:20:05) Mihai Moldovan: so reducing the number of components is currently not a high priority or even easily possible
(15:20:16) ***Mike#1 agrees on that.
(15:20:26) Mihai Moldovan: I'm continuing now with x2goclient
(15:20:35) Alex: ok, we just not touching them :)
(15:21:03) Mike#1: (do we finish at 15.00 UTC?)
(15:21:09) Mihai Moldovan: as recently discussed on the ML, I want to merge my feature/cleanup branch into master shortly
(15:21:20) Mike#1: +1 from me!!!
(15:21:30) Mihai Moldovan: a visual representation is here: http://code.x2go.org/x2goclient?p=x2goclient.git;a=tree > http://code.x2go.org/x2goclient?p=x2goclient.git;a=tree;h=refs/heads/feature/cleanup;hb=refs/heads/feature/cleanup
(15:21:48) Mihai Moldovan: so far, I've tested OS X builds and help the branch updated with master
(15:22:07) Mihai Moldovan: so windows and linux should still equally work, but will need at least one round of testing
(15:22:34) Mihai Moldovan: yesterday I've fixed the maemo/n900 project file to meet the new directory structure
(15:23:07) Heinz-M. Graesing: Ionic: we should create a wiki page for "public" test (if we don't have one already)
(15:23:15) Mihai Moldovan: when it comes to maemo/n900, as I've discussed with Heinz privately, we don't have active testing of that feature and should think about removing it after 4.0.4.0
(15:23:30) Heinz-M. Graesing: with our bugtracker linked/a ready to use form for our bugtracker
(15:23:37) Alex: Is someone still using maemo?
(15:23:45) Mihai Moldovan: we have genuinely no idea
(15:23:51) Heinz-M. Graesing: Ionic: Igot my N900 up and running again :)
(15:24:07) Mihai Moldovan: Heinz told me there some N900 lovers around, but I doubt there's high demand
(15:24:07) Heinz-M. Graesing: Alex|2: yes - there is even a new N900 device...
(15:24:23) Mihai Moldovan: and I feel bad shipping something that I have no idea if it works
(15:24:27) Heinz-M. Graesing: Alex|2: N900 without the Nokia Logo
(15:24:52) Heinz-M. Graesing: Alex|2: I reassembled one using 3 non working devices...
(15:25:13) Mihai Moldovan: I will need access to the Windows X2Go build VM
(15:25:20) ***Mike#1 waves with his N900, also...
(15:25:29) Mihai Moldovan: sunweaver: are you using x2goclient on N900?
(15:25:33) Mike#1: mikedep333: can you do that for Ionic?
(15:25:36) Mihai Moldovan: who should I bug with that?
(15:25:41) Mike#2: yes
(15:25:42) Mike#1: I used to, but in a Debian chroot.
(15:26:00) Mihai Moldovan: ok, great
(15:26:17) Mihai Moldovan: I really need to test my branch for basic compilation on Windows and Linux
(15:26:22) Mihai Moldovan: once that's done, I'll merge
(15:26:45) Heinz-M. Graesing: http://neo900.org/
(15:27:06) Mihai Moldovan: next up (to ramp up the pace): code.x2go.org/git
(15:27:28) Mihai Moldovan: so far, I've only been able to rewrite the hooks to allow for the new policy I announced some months ago
(15:27:44) Mihai Moldovan: I had to fix a bug yesterday, but it seems to be working now
(15:28:11) Mihai Moldovan: I still need to put 2 git maintenance scripts (non-hooks) into the new maintenancescripts.git repository, which is now also public
(15:28:23) Mihai Moldovan: (formerly our hooks were non-published)
(15:28:24) ***Mike#1 just did a test-build of x2goclient from feature/cleanup against a clean Debian unstable chroot. Looks good.
(15:28:50) Mike#1: x2goclient should be bumped to 4.0.4.0 after the merge.
(15:28:58) Mihai Moldovan: what is bothering me a lot more is the git access model
(15:29:07) Mihai Moldovan: sunweaver: yes, master and feature/cleanup already reference 4.0.4.0
(15:29:35) Mihai Moldovan: currently, anyone with access to git@code.x2go.org has access to the git repositories
(15:29:39) Mihai Moldovan: AND TO A SHELL!
(15:29:54) Mihai Moldovan: so, in effect, *everyone* can add new SSH keys
(15:30:03) Mihai Moldovan: to that account only, but I find that bad enough
(15:30:14) Mihai Moldovan: what I want to do is:
(15:30:21) Heinz-M. Graesing: Ionic: we should change that...
(15:30:30) Heinz-M. Graesing: sry...
(15:30:32) Mihai Moldovan:   - make that file unwritable by the x2go user (chmod +i)
(15:31:07) Mihai Moldovan:   - create a script that can be run by x2go-admin via sudo, which changes x2go's .authorized_keys file
(15:31:51) Mihai Moldovan:   - keep SSH keys in a new directory in x2go-admin's account, one directory per person with one file per SSH key
(15:32:22) Mihai Moldovan: those files will need to be concatenated by the script and x2go's .authorized_keys file replaced with the concatenated result
(15:32:37) Mike#1: sounds good.
(15:32:46) Mihai Moldovan: x2go's .authorized_keys file should always remain unwritable to by any user (especially x2go)
(15:33:05) Mihai Moldovan: kudos to MyNameIsRetro for suggesting chmod -i/+i
(15:33:44) Mihai Moldovan: (other ways to make the file unreadable other than setting it immutable is not working, as opensshd will deny using it, if the permissions don't match exactly what it expects)
(15:34:05) Mihai Moldovan: maybe immutable will make openssh reject it, too, I'll have to try that out locally on my box
(15:34:41) Mihai Moldovan: in the long run, I'd also like remove unused SSH keys based on the last login time
(15:35:01) aaronr hat den Raum verlassen (quit: Remote host closed the connection).
(15:35:05) Mike#2: selinux would do it! (Sorry, I had to point that out.)
(15:35:10) Mihai Moldovan: if, say, the SSH key hasn't been used in 1 year, mark it for deletion, notify the owner and some time after that, disable it for good
(15:35:41) Mihai Moldovan: this is Priority: Wishlist, and I have no idea how that can be implemented currently
(15:36:01) Mihai Moldovan: mikedep333: selinux would do what? immutable .authorized_keys file?
(15:36:10) Mihai Moldovan: or reject the file?
(15:36:25) Mike#2: selinux would let you make the file read-only by its owner
(15:36:41) Mihai Moldovan: I'm pretty sure we could also achieve that via linux ACL's
(15:36:57) Mihai Moldovan: (if Mike's VM is using an ACL-capable file system)
(15:37:06) Heinz-M. Graesing: Ionic: are ACL's already activated?
(15:37:16) Mihai Moldovan: h1Org: file system ACL.
(15:37:20) Mihai Moldovan: h1Org: not git ACL.
(15:37:53) Mihai Moldovan: the ACL git hooks are not activated and I don't plan on doing that any time soon
(15:37:54) Heinz-M. Graesing: Ionic: yes - but they can be confusing to, as there is no "visualizing" by "normal" commands
(15:38:15) Mihai Moldovan: h1Org: I am aware, this needs to be documented
(15:38:28) Heinz-M. Graesing: Ionic: so maybe the info shouls be stored in /etc/issue or modt
(15:38:34) Heinz-M. Graesing: should
(15:39:04) Mihai Moldovan: comments/questions?
(15:39:36) Mihai Moldovan: (especially from Alex|2, as he seemed interested last time around)
(15:40:19) Alex: sounds ok for me
(15:42:16) Kjetil Fleten hat den Raum verlassen (quit: Quit: kfleten).
(15:42:58) Mihai Moldovan: the list of people who currently have access to pure git are: Alex, Heinz, Mike#1, Mike#2, Jan, Orion, Reinhard, Moritz, me and two other keys I have no idea who they belong to
(15:43:18) Mihai Moldovan: "snalwuer" and "teranders"
(15:43:39) Henning Heinold hat den Raum verlassen (quit: Ping timeout: 240 seconds).
(15:44:58) Mike#1: snalwuer -> Alexander Wuerstlein, teranders (aka tan here on IRC), Terje Anderson (former i18n team leader).
(15:45:27) Mihai Moldovan: people with access to x2go-admin: Alex, Heinz, Mike#1, Mike#2, Reinhard, Clemens, me
(15:45:45) Mihai Moldovan: the x2go-admin account is currently used for git administration and manual package upload
(15:45:49) Mike#2: Ionic, and "borisb" and "natashaf"
(15:45:50) Mike#1: basically, x2go-admin people are those that can put stuff under code.x2go.org/releases/
(15:45:51) Mike#2: jk
(15:45:54) Alex: Who from this people still doing something for x2go?
(15:46:15) Mike#1: Drop Reinhard.
(15:46:19) Mihai Moldovan: sunweaver: well, it's also the account who has access to git admin stuff...
(15:46:29) Mihai Moldovan: which likewise is a little bit interweaved
(15:46:33) Mike#1: yeah, maybe this should be separated.
(15:46:52) Mihai Moldovan: it's not about trust, but damage reduction
(15:47:02) Mihai Moldovan: in case of shit -> fan
(15:47:21) Mike#1: Drop Terje and Alexander Wuerstlein and also Orion (he never used the commit access).
(15:47:30) Mihai Moldovan: ok
(15:47:42) Alex: yes, drop every one, who not working active for x2go. It is easy to insert them again, if it needed.
(15:47:49) Mike#1: also Jan, actually.
(15:47:56) Mike#2: and update
(15:47:56) Mike#2: http://wiki.x2go.org/doku.php/doc:organigram
(15:48:19) Mike#1: and also Moritz.
(15:48:39) Mihai Moldovan: done
(15:48:44) Mihai Moldovan: hum
(15:48:46) Mihai Moldovan: ok
(15:48:54) Mihai Moldovan: aren't they contributing anymore?
(15:49:07) Mihai Moldovan: (curious about what happened here)
(15:50:05) Mihai Moldovan: ok, all done
(15:50:17) Mihai Moldovan: will update the wiki page later on
(15:51:08) Mike#1: Terje would like to rejoin, but was busy.
(15:51:24) Mike#1: Jan never used his commit access, but is still active under the hood of Michael Kromer, I guess.
(15:51:35) Mike#1: Orion is active, just not upstream.
(15:51:44) steveeJ [~junky@virthost3.stefanjunker.de] hat den Raum betreten.
(15:51:45) Mike#1: Reinhard left $UNI and is in the US now.
(15:52:03) Mihai Moldovan: that's also on my todo list...
(15:52:04) Mihai Moldovan: scnr
(15:52:18) Mike#1: also drop Morty, actually. (still around, but not active, anymore).
(15:52:42) Mihai Moldovan: I did
(15:52:59) Mihai Moldovan: Morty = Moritz
(15:53:04) Mike#1: ye
(15:53:05) Mike#1: s
(15:53:20) Mike#1: ok, ten minutes to go...
(15:53:23) Mike#1: Ionic?
(15:53:30) Mihai Moldovan: I guess I'm done
(15:53:49) Mike#1: h1Org: ?
(15:54:56) Heinz-M. Graesing: sry - AWK
(15:55:23) Mihai Moldovan: gawk? mawk?
(15:55:39) Mike#1: afk.
(15:56:01) Mike#1: Ionic: my feature/cleanup branch was not up-to-date, .deb builds fail, actually.
(15:56:04) Heinz-M. Graesing: Ionic: I needed to go away from my keyboard/table/room to get a backup device
(15:56:15) Mihai Moldovan: sunweaver: oh, ok
(15:56:15) Mike#1: Short Introduction about where to find information about "Package Building"
(15:56:26) Mike#1: http://wiki.x2go.org/doku.php/wiki:development:build-server
(15:56:30) Mike#1: needs love...
(15:56:47) Mihai Moldovan: yes, also the client stuff, especially on OS X
(15:57:04) Mihai Moldovan: the current information is outdated for features/cleanup
(15:57:13) Mihai Moldovan: OS X is handled by macbuild.sh and its options now
(15:57:22) Mihai Moldovan: (sunweaver: run ./macbuild.sh --help)
(15:58:03) Heinz-M. Graesing: sunweaver: my key is already dropped due an security issue
(15:58:25) Mike#1: ?
(15:58:35) Mike#1: send a new pubkey to Ionic then.
(15:58:39) Mihai Moldovan: there are quite some keys for you in the authorized_keys file
(15:58:39) Mike#1: SSH key?
(15:58:45) Mihai Moldovan: file(s)
(15:58:51) Heinz-M. Graesing: sunweaver: yes
(15:58:51) Mike#1: next: Which bugs have been closed with the last update?
(15:59:00) Mike#1: the closed bugs can be found
(15:59:04) Mihai Moldovan: I'll send you all your current keys and ask what can be kept and what removed
(15:59:05) Mike#1: (a) in the release announcements.
(15:59:16) Mike#1: (b) on x2go-dev.
(15:59:32) Mike#1: no need to go through them here, it is fetchable information to those who are interested.
(16:00:09) Mike#1: also, everyone can check debian/changelog files on each X2Go component. E.g. http://code.x2go.org/gitweb?p=x2goclient.git;a=blob;f=debian/changelog;h=52c0157ff453060798313e7c99e8f962b68c33c2;hb=3365117e4b0f0fed12e3d0812bb1647438873552
(16:00:38) Heinz-M. Graesing: sunweaver: we should think about a statement on the blog, since this time there were some really important bugs closed...
(16:01:03) Mike#1: I won"t continue feeding that blog.
(16:01:30) Mike#1: but I agree with you, that we should present some more human readable text somewhere.
(16:02:15) Mike#1: I plan to use my blog more actively on all the FLOSS work I do, so I"d be happy having a planet.x2go.org site instead of a real blog.
(16:02:37) Mike#1: so that people can run their own blogs (or even an X2Go team blog)...
(16:03:24) Mike#1: ok. Time is over, it seems.
(16:03:50) Mike#2: I will discuss VcXsrv via the ML: http://wiki.x2go.org/doku.php/2015-03:day-2015-03-05?s[]=meeting
(16:04:10) Mihai Moldovan: mikedep333: we can do it on the next meeting, too
(16:04:20) Mike#2: I do not wish to wait 2 weeks.
(16:04:28) ***Mike#1 nods...
(16:04:37) Mike#1: use the ML and present what you have in mind.
(16:04:45) Mike#2: will do
(16:04:52) Heinz-M. Graesing: mikedep333: please fill out our query on: http://wiki.x2go.org/doku.php/wiki:development:planning:start
(16:05:00) Mike#2: ok
(16:05:03) Mihai Moldovan: ok
(16:05:17) Heinz-M. Graesing: I'll try my best to be available as soon as possible!
(16:05:32) Mike#2: thanks!
(16:05:39) Heinz-M. Graesing: -> section Next ΞX2Go Development Meeting
(16:05:57) Heinz-M. Graesing: Thank you very much for taking part!!!
(16:06:21) Mike#1: thanks h1Org for getting us at the table.
(16:06:46) Mike#1: everyone, please remember your todos and keep to agreed schedules on assigned tasks. Thanks.
(16:07:10) Mike#1: time-of-day for today was optimal from my point of view.
(16:07:26) Mike#2: hmm
(16:07:36) Mike#1: not?
(16:07:41) Mike#2: were there 2 different links for today's meeting?
(16:07:44) Mike#2: http://wiki.x2go.org/doku.php/2015-03:day-2015-03-05?s[]=meeting
(16:08:02) Mihai Moldovan: no, just one, I guess?
(16:08:09) Mike#2: ok
(16:08:14) Mike#1: I used the one above.
(16:08:38) Mihai Moldovan: it's the same link, isn't it?
(16:08:41) Mike#2: I will mark myself as available for all 3 dates
(16:08:47) Mike#2: but it depends on the time of day
(16:08:51) ***Stefan Baur does not agree on the optimal time of day statement by sunweaver
(16:09:01) Mike#1: ################################################## official ending of the meeting ####################
2015-03/day-2015-03-05.txt · Last modified: 2015/03/06 21:50 by stefanbaur