Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2014-03-03

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:55 bala joined #gluster-dev
02:59 * jclift yawns
03:19 bharata-rao joined #gluster-dev
03:57 ira joined #gluster-dev
04:04 aravindavk joined #gluster-dev
04:08 bala joined #gluster-dev
04:19 hagarth joined #gluster-dev
04:22 mohankumar__ joined #gluster-dev
04:24 ndarshan joined #gluster-dev
04:33 shubhendu joined #gluster-dev
04:34 mohankumar__ joined #gluster-dev
04:35 ppai joined #gluster-dev
04:47 kaushal_ joined #gluster-dev
04:48 mohankumar__ joined #gluster-dev
04:56 mohankumar__ joined #gluster-dev
05:05 mohankumar__ joined #gluster-dev
05:13 mohankumar__ joined #gluster-dev
05:26 aravindavk joined #gluster-dev
05:26 mohankumar__ joined #gluster-dev
05:42 deepakcs joined #gluster-dev
05:49 lalatenduM joined #gluster-dev
05:51 hagarth joined #gluster-dev
05:53 ajha joined #gluster-dev
05:55 kdhananjay joined #gluster-dev
06:15 raghu` joined #gluster-dev
06:26 hagarth joined #gluster-dev
07:00 hagarth joined #gluster-dev
07:04 aravindavk joined #gluster-dev
07:38 deepakcs joined #gluster-dev
07:48 hagarth joined #gluster-dev
08:03 aravindavk joined #gluster-dev
08:15 ndevos hagarth: I've created a 3.5.1 tracker: https://bugzilla.redhat.com/showdep​endencytree.cgi?id=glusterfs-3.5.1
08:20 ndevos jclift: I can not edit any wiki pages under http://www.gluster.org/community/documentation/ , you blocked someone by IP? "Your current IP address is 127.0.0.1, and the block ID is #513."
08:45 lalatenduM joined #gluster-dev
09:12 hagarth joined #gluster-dev
09:20 aravinda_ joined #gluster-dev
09:58 aravinda_ joined #gluster-dev
10:06 bala joined #gluster-dev
10:07 hagarth joined #gluster-dev
10:48 aravinda_ joined #gluster-dev
11:06 kanagaraj joined #gluster-dev
11:11 jclift ndevos: Ugh.  Yeah I nuked some spammer, and also checked the "block this IP address" option too
11:12 hagarth joined #gluster-dev
11:12 jclift ndevos: Have you fixed the block, or should I go unblock the IP?
11:20 jclift ndevos: IP address removed from the block list
11:24 shyam joined #gluster-dev
11:39 aravindavk joined #gluster-dev
11:45 hchiramm__ joined #gluster-dev
12:26 hagarth joined #gluster-dev
12:30 edward1 joined #gluster-dev
12:41 bala joined #gluster-dev
12:45 kkeithley jclift: Re: lvm2-devel rpm. Yeah, that's too bad, but c'est la vie. But since it's el5...  If anyone really wants to use glupy then they just have to do it on >= el6 or fedora, &c.
12:53 jclift kkeithley: ?
12:54 jclift kkeithley: The thing that's not working on EL5 is geo-rep.
12:54 * jclift hasn't tested Glupy on EL5 yet
12:54 jclift Guess I really should though. :)
12:54 kkeithley okay. because old python
12:54 jclift Yeah
12:55 jclift It'll look at Glupy on EL5 after I take a look at the regression failure on glupy.t later
12:55 kkeithley and we can't use the python26 rpms?
12:55 * jclift didn't know there are python 2.6 rpms
12:56 jclift I was just using the python that comes with EL5.
12:56 jclift I don't have much/any experience with Python on EL5.
12:56 kkeithley yup, the default, out-of-the-box, python package is 2.4.  I didn't know either but I just guessed and checked
12:56 jclift Interesting.
12:57 jclift Maybe we should add some kind of check to the startup of geo-rep, to ensure it's a supported version of Python, and exit with nice warning if it's using the old Python version.
12:59 kkeithley If we have BuildRequires: python26(-devel) on el5 and then either an explicit or magic-autodep Requires: python26 then that's probably all the checking we need?
13:00 kkeithley ndevos: ^^^
13:00 jclift That might be the way to go. :)
13:01 ndevos kkeithley: BuildRequires on python26-devel for sure, http://fedoraproject.org/wiki/Packaging:Python demands that
13:01 jclift ndevos: Cool.
13:01 jclift ndevos: It turns out geo-rep on EL5 won't start, due to the Python 2.4.x that comes with EL5 by default.
13:02 jclift ndevos: Sounds like this will fix that prob. :)
13:03 jclift kkeithley: Hmmm, since I have the CentOS 5 VM handy, I'll try out a patch with that right now and see if it fixes things.
13:10 bala joined #gluster-dev
13:12 ndevos jclift, kkeithley: ah, python26-devel? not sure if python26 is available by default on RHEL5...
13:12 kkeithley I see there's a python26-simplejson, but no python26-ctypes
13:13 kkeithley it's probably in epel
13:13 kkeithley or I'm sure it's in epel
13:13 kkeithley which doesn't bother me for community rpms
13:13 kkeithley unless you've got a good argument otherwise
13:13 ndevos no, thats okay
13:15 kkeithley koji builds can use epel packages, right?
13:15 kkeithley oh, I know they can
13:16 kkeithley that is to say, koji builds of epel rpms can use other epel rpms that are already built in koji
13:18 jclift ndevos: Yeah.  By default (on a basic minimal type install I think) it was only Python 2.4.  But there are python26 and python26-devel rpms in EPEL.
13:18 jclift As kkeithley mentions, no python-ctypes though.
13:19 jclift No, I'm wrong, it's definitely in epel 5.
13:19 kkeithley not as a separate package, but I just built with python26-devel and `rpm -q --requires -p glusterfs-geo-replication-3g​it-1.el5.centos.x86_64.rpm` gives ... python26  python26-ctypes ...
13:19 jclift Ahhh no.  It's python-ctypes, not python26-ctypes.
13:20 ndevos maybe python26-ctypes is part of python26(-devel)?
13:20 kkeithley maybe, I'm looking to see where it got it from
13:20 jclift Don't ask me.  I'm obviously newbie wwith this stuff. ;)
13:22 jclift Interestingly, after tweeting yesterday about the new EL5 rpms instructions on the wiki, the page count for the compilingrpms page jumped by over 100.
13:23 jclift So it looks like at least some people found it interesting.
13:23 kkeithley oop, I made a mistake
13:23 jclift ?
13:24 kkeithley I fubared. rebuilding....
13:24 jclift Does this look right? http://fpaste.org/81831/85307313/
13:25 jclift Using a Requires of python26, and the buildrequires of python26-devel.
13:25 jclift But conditional for EL5 and below only
13:25 * jclift figures EL4 isn't supported
13:25 kkeithley We haven't ever had el4 builds
13:25 jclift :)
13:26 jclift No desire to start now here. :)
13:26 kkeithley as ndevos points out and I see empirically, the Requires is automatically determined by rpmbuild autodep magic. You don't need the Requires: python26
13:27 jclift k
13:28 jclift Doesn't seem to work in practise though.
13:28 jclift Not the Requires automagic.
13:29 jclift I mean the fpaste approach.
13:29 jclift fpasted approach
13:29 jclift kkeithley: Installing rpms generated with the python26 dependency errors: http://fpaste.org/81832/53351139/
13:30 jclift I seems like it's still trying to call Python 2.4.
13:31 ndevos jclift: yeah, I expect you need to fix the #!/usr/bin/python for the 2.6 scripts too
13:31 jclift ndevos: yeah, was just writing that.
13:31 jclift Maybe some magic to do with configure scripting or autoconf?
13:32 jclift Either way, this has gone from a simple fix to something more complicated.
13:32 jclift Probably better if I just write up the detailed by report and let Venky figure out if we want to fix it or not, etc.
13:32 jclift s/by report/bug report/
13:33 jclift Thanks ndevos kkeithley. :)
13:34 ndevos jclift: yeah, file a bug for that
13:34 ndevos @file a bug
13:34 * ndevos kicks glusterbot
13:34 jclift It's ok.  I'm already on BZ. :)
13:34 kkeithley I don't think glusterbot works here in #gluster-dev
13:34 ndevos glusterbot: ping
13:34 glusterbot pong
13:35 ndevos glusterbot: @file a bug
13:35 kkeithley my patch looks like this  http://paste.fedoraproject.org/81835/13938536
13:36 ndevos kkeithley: doesnt fedora have python-2.7?
13:36 jclift kkeithley: Guessing that's strategically placed defines for only affecting EL5?
13:36 kkeithley I'm not sure why we have explicit Requires: python{,26} in %package geo-rep (and %package regression).
13:36 ndevos maybe 'python >= 2.6'?
13:36 jclift F19 definitely has 2.7.x
13:37 kkeithley ndevos: yes, it has -2.7, but having BuildRequires: python >= 2.6 won't get you the python26-* rpms on el5 AFAIK
13:38 jclift Good point
13:39 jclift Yeah, on F19 the "python" package is Python 2.7.5.
13:39 ndevos kkeithley: why not? I'd guess it should pull in python26, assuming pyhtin26 "Provides: python = 2.6"
13:39 * jclift just checked by firing up an old F19 VM
13:39 kkeithley Let's try and see?
13:40 ndevos jclift@centos510: ! rpm -q --provides python26
13:41 jclift "package python26 is not installed"
13:41 jclift That's on F19 with the normal python package installed
13:42 ndevos jclift: you missed the @centos510 part?
13:42 kkeithley on my centos510 vm, with python26 installed.   % configure; make ....
13:42 jclift ndevos: Nah, run that afterwards.
13:42 jclift Took a bit to cut-n-paste.
13:42 jclift http://fpaste.org/81837/39385414/
13:42 jclift ndevos: ^^^
13:43 jclift But yeah, missed it at first. :)
13:43 kkeithley doesnt work.. on my centos510 vm, with python26 installed.   % configure; make .... I get http://paste.fedoraproject.org/81838/54171139
13:43 kkeithley
13:44 kkeithley python26 installed but BuildRequires: python >= 2.6
13:44 ndevos kkeithley: you have python26-devel installed?
13:44 kkeithley I do
13:45 kkeithley rpm -q python26-devel
13:45 kkeithley python26-devel-2.6.8-2.el5
13:45 kkeithley python26-devel-2.6.8-2.el5
13:45 jclift Are we better to let Venky figure this out later, or should we keep plugging away at this now?
13:45 jclift Regardless of how we get these deps sorted, we're still going to have the issue of picking the runtime python properly. :?
13:46 jclift :/
13:46 ndevos kkeithley: "rpm -q --provides python26-devel" ?
14:07 jclift Ugh.
14:09 jclift The feeling you get when your supposedly reputable domain registrar (hover.com) deletes a support ticket after issues yesterday, their support system ignores all future requeests, and their online support system "has a problem" when trying to submit a ticket that way.
14:11 kkeithley ndevos: yes, `rpm -q --provides python26-devel` gives python26-devel = 2.6.8-2.el5, but I still have http://paste.fedoraproject.org/81838/54171139 anyway.
14:11 kkeithley Do you get something different?
14:30 shyam joined #gluster-dev
14:40 ndevos kkeithley: atm I don't have a centos 5.10 to test
14:40 ndevos kkeithley: does it maybe also provide python2-devel = 2.6 ?
14:42 an_ joined #gluster-dev
15:04 lpabon joined #gluster-dev
15:05 kkeithley I only see python26-devel
15:06 kkeithley or to be clearer, this is what I see: http://paste.fedoraproject.org/81875/59201139
15:14 ndevos hmm, I guess there is a reason for that, but it doesnt help in this case
15:18 kkeithley I'm not sure. If both python-devel and python26-devel provided python2-devel, that seems like it'd be a conflict. To the extent that we want geo-rep to work on crufty old el5 then I suspect we have to explicitly require python26{,-devel}.
15:19 kkeithley What worries me though is that I don't know how well it works to have both python and python26 on the box. Because I don't see an easy way to remove python and leave python26. Too much other stuff depends on python
15:20 kkeithley And I don't know what magic it would take to make geo-rep unambiguously use python26.
15:24 ndk joined #gluster-dev
15:26 kkeithley e.g. change all the #!/usr/bin/env python -> #!/usr/bin/env python26.    yuck.
15:31 jobewan joined #gluster-dev
16:09 an_ joined #gluster-dev
16:32 jclift kkeithley: Any idea how feasible it would be to do the python -> python26 thin using autoconf?  eg making gsynd be gsynd.in, and using a replacable symbol on the #!/usr/bin/env __PYTHON_VERSION__ line?
16:32 jclift It'd have to only be triggered for EL5 though.  Not sure if autoconf can do conditional things that way.
16:37 kanagaraj joined #gluster-dev
16:41 kkeithley I was thinking more along the lines of running *.py through a sed script to change #!/usr/bin/env python to #!/usr/bin/env python26; on el5 only. Simple enough that it should not need a replaceable symbol. And not require editing all the files.
16:43 kkeithley well, sorry, it's not a whole lot different than what you're proposing, but I think it's simple enough to not require editing every file to change python to __PYTHON_VERSION__
16:43 kkeithley And not autoconf, but many a %postin scriptlet in the spec file.
16:44 kkeithley I'm not convinced though that anything along these lines isn't a horrible hack, and not just a horrible hack but a horrible hack for a kind of corner case, little used at this point version of RHEL.
16:59 shruti joined #gluster-dev
16:59 shruti left #gluster-dev
17:00 an_ joined #gluster-dev
17:11 jclift kkeithley: Yeah, I'm ambivalent about it.
17:25 kanagaraj joined #gluster-dev
17:52 edward1 joined #gluster-dev
17:55 edward1 joined #gluster-dev
17:59 shyam left #gluster-dev
18:38 an_ joined #gluster-dev
21:04 ndevos kkeithley, jclift: I am not opposed to dropping geo-rep for EL5, maybe one of you can send an email to the gluster-users@ and gluster-devel@ to get some feedback?
21:05 ndevos it would be stupid to spend time on trying to fix something that nobody would use anyway...
22:45 jclift ndevos: I've emailed Venky (geo-rep maintainer), describing the problem in some detail, asking for his thoughts on it, asking if he's aware of demand level, and ask mentioned we're thinking that it might be a good idea to ask on the mailing lists.
22:46 jclift ndevos kkeithley: Prob hear back from him tomorrow, etc.

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary