Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2016-08-05

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

All times shown according to UTC.

Time Nick Message
00:41 inevity joined #gluster-dev
00:50 inevity joined #gluster-dev
01:24 shyam joined #gluster-dev
01:33 shyam left #gluster-dev
01:39 hagarth joined #gluster-dev
02:19 spalai joined #gluster-dev
02:28 julim joined #gluster-dev
03:11 magrawal joined #gluster-dev
03:28 skoduri joined #gluster-dev
03:44 nbalacha joined #gluster-dev
03:49 atinm joined #gluster-dev
03:50 nishanth joined #gluster-dev
04:06 spalai joined #gluster-dev
04:11 inevity joined #gluster-dev
04:13 spalai joined #gluster-dev
04:24 ira joined #gluster-dev
04:30 itisravi joined #gluster-dev
04:34 shubhendu joined #gluster-dev
04:37 poornimag joined #gluster-dev
04:38 ankitraj joined #gluster-dev
04:43 aspandey joined #gluster-dev
04:44 Bhaskarakiran joined #gluster-dev
04:50 kdhananjay joined #gluster-dev
04:50 sanoj joined #gluster-dev
04:54 itisravi joined #gluster-dev
04:55 Apeksha joined #gluster-dev
05:03 ppai joined #gluster-dev
05:03 ankitraj joined #gluster-dev
05:06 pranithk1 joined #gluster-dev
05:12 karthik_ joined #gluster-dev
05:19 jiffin joined #gluster-dev
05:24 Muthu joined #gluster-dev
05:25 spalai joined #gluster-dev
05:25 aravindavk joined #gluster-dev
05:26 ndarshan joined #gluster-dev
05:33 jiffin1 joined #gluster-dev
05:36 hgowtham joined #gluster-dev
05:36 Manikandan joined #gluster-dev
05:36 aravindavk joined #gluster-dev
05:40 mchangir joined #gluster-dev
05:45 poornimag joined #gluster-dev
05:45 aspandey joined #gluster-dev
05:46 rafi joined #gluster-dev
05:48 karthik_ joined #gluster-dev
05:49 ppai joined #gluster-dev
05:51 asengupt joined #gluster-dev
05:54 rafi1 joined #gluster-dev
06:16 mchangir joined #gluster-dev
06:26 nishanth joined #gluster-dev
06:26 rastar joined #gluster-dev
06:32 asengupt joined #gluster-dev
06:36 hchiramm joined #gluster-dev
06:49 rastar joined #gluster-dev
06:57 post-factum The GNU C Library version 2.24 is now available https://lwn.net/Articles/696469/
06:57 post-factum The readdir_r and readdir64_r functions have been deprecated.  It is
06:57 post-factum recommended to use readdir and readdir64 instead.
07:05 jiffin joined #gluster-dev
07:06 msvbhat joined #gluster-dev
07:07 anoopcs And we already have those changes merged in master.
07:07 anoopcs https://review.gluster.org/#/c/14838/
07:08 rafi1 joined #gluster-dev
07:09 aravindavk joined #gluster-dev
07:10 devyani7_ joined #gluster-dev
07:13 devyani7_ joined #gluster-dev
07:18 Manikandan joined #gluster-dev
07:18 ashiq joined #gluster-dev
07:22 ppai joined #gluster-dev
07:23 karthik_ joined #gluster-dev
07:24 post-factum yay then :)
07:26 atalur joined #gluster-dev
07:28 aspandey joined #gluster-dev
07:30 poornimag joined #gluster-dev
07:45 suliba joined #gluster-dev
07:50 mchangir joined #gluster-dev
07:50 Apeksha_ joined #gluster-dev
07:53 Bhaskarakiran joined #gluster-dev
07:55 Bhaskarakiran joined #gluster-dev
08:00 ndevos nigelb: could you can clear the Verified=-1 on http://review.gluster.org/14955
08:00 ndevos ?
08:01 nigelb yep
08:02 nigelb ndevos: done.
08:02 ndevos nigelb++ thanks
08:02 glusterbot ndevos: nigelb's karma is now 20
08:03 nigelb I'm sort of writing out my plans for the maintainer check for patches
08:03 nigelb https://public.pad.fsfe.org​/p/gluster-check-maintainer
08:03 nigelb Feel free to chime in there
08:04 ndevos ok, I'll check that later
08:05 nigelb Not writing code yet. Just thinking out loud.
08:13 ashiq joined #gluster-dev
08:18 hchiramm joined #gluster-dev
08:33 jiffin rjoseph: i had addressed ur previous comments on
08:33 jiffin http://review.gluster.org/#/c/13763/
08:33 jiffin http://review.gluster.org/#/c/13764/
08:33 jiffin rjoseph: can you please take a look again
08:58 Apeksha joined #gluster-dev
09:03 Manikandan joined #gluster-dev
09:10 spalai ndevos++
09:10 glusterbot spalai: ndevos's karma is now 297
09:16 atalur joined #gluster-dev
09:29 mchangir atinm, there are two volume options with the same name: performance.cache-size     defined in xlators/mgmt/glusterd/src/glusterd-volume-set.c ... is that intentional?
09:32 atinm mchangir, this is a known problem, we tried to address it through http://review.gluster.org/#/c/10262 but unfortunately this still has some loopholes on backward compatibility
09:34 mchangir atinm, ok
09:38 msvbhat joined #gluster-dev
09:48 penguinRaider joined #gluster-dev
09:55 nigelb rjoseph: do you still need a netbsd machine now that you know what's going wrong?
10:07 atalur joined #gluster-dev
10:12 mchangir joined #gluster-dev
10:27 ankitraj joined #gluster-dev
10:32 ppai ndevos, Hi, do you maintain glusterfs centos6 packages in the storage SIG ?
10:36 ppai on centos6, installing gluster requires epel repo too as pyxattr package (needed by glusterfs-server) isn't shipped in base or gluster repo. http://paste.openstack.org/show/550386/ Should this be documented somehwere ?
10:37 ndevos ppai: pyxattr is packaged in the CentOS Storage SIG, epel should not be needed
10:37 ndevos ppai: packages are mostly just rebuilds from Fedora, and the .specs are kept under https://github.com/CentOS-Storage-SIG
10:38 ndevos ppai: on a CentOS-6, you should be able to do: yum install centos-release-gluster && yum install glusterfs-server
10:39 ndevos ppai: the testing repository is at http://buildlogs.centos.org/cent​os/6/storage/x86_64/gluster-3.8/
10:39 karthik_ joined #gluster-dev
10:39 ndevos ppai: and the mirrored repo is at http://mirror.centos.org/centos​/6/storage/x86_64/gluster-3.8/
10:39 ppai ndevos: May be I'm doing something wrong then. I got a brand new VM from "centos 6.8 minimal" iso. And seems like pyxattr isn't there unless I added EPEL. http://paste.openstack.org/show/550390/
10:40 ndevos ppai: hmm, and pyxattr is missing in the 2nd repo...
10:40 ndevos ppai: you should be able to install pyxattr with "yum --enablerepo=centos-gluster38-test install pyxattr"
10:41 * ndevos checks why it isnt on the mirrors
10:41 ppai I see
10:42 ndevos hmm, it has been tagged as -release, meaning the issue is with the centos admins
10:46 ndevos ppai: installing centos-release-gluster should pull in centos-release-storage-common too, can you check if that got installed?
10:46 ndevos ppai: other than that, it looks ok, and I'll get the missing pyxattr issue resolved asap
10:48 ppai ndevos: centos-release-storage-common is in there
10:48 suliba joined #gluster-dev
10:48 ndevos ok
10:48 ppai ndevos: except for pyxattr, nothing else is pulled from EPEL.
10:49 ndevos ppai: yeah, none of the SIGs should pull in anything from non-CentOS repositories... I wonder how it can happen that pyxattr wasnt signed+pushed
10:49 ppai ndevos++
10:49 glusterbot ppai: ndevos's karma is now 298
10:50 ndevos ppai++ thanks for reporting!
10:50 glusterbot ndevos: ppai's karma is now 15
11:03 msvbhat joined #gluster-dev
11:04 ndevos sac++ for https://bugzilla.redhat.co​m/show_bug.cgi?id=1344276
11:04 glusterbot ndevos: sac's karma is now 2
11:04 glusterbot Bug 1344276: medium, unspecified, ---, nobody, NEW , Review Request: gdeploy - Tool to deploy GlusterFS clusters and other utilities
11:04 ndevos misc: now is your chance! maybe you want to review ^ and sponsor sac?
11:05 pur joined #gluster-dev
11:08 rjoseph nigelb: I don't need the machine as of now. Thanks
11:08 inevity joined #gluster-dev
11:08 nigelb cheers, I'll close the bug.
11:09 Apeksha joined #gluster-dev
11:09 sac ndevos, thanks :-)
11:15 ramky joined #gluster-dev
11:22 hchiramm joined #gluster-dev
11:23 kkeithley excellent sac++ for https://bugzilla.redhat.co​m/show_bug.cgi?id=1344276
11:23 glusterbot kkeithley: sac's karma is now 3
11:23 glusterbot Bug 1344276: medium, unspecified, ---, nobody, NEW , Review Request: gdeploy - Tool to deploy GlusterFS clusters and other utilities
11:26 penguinRaider joined #gluster-dev
11:27 aspandey xavih, ping
11:27 glusterbot aspandey: Please don't naked ping. http://blogs.gnome.org/mark​mc/2014/02/20/naked-pings/
11:27 kkeithley glusterbot--
11:27 glusterbot kkeithley: glusterbot's karma is now 1
11:36 Saravanakmr joined #gluster-dev
11:44 rraja joined #gluster-dev
11:51 ramky joined #gluster-dev
11:54 Manikandan joined #gluster-dev
12:12 julim joined #gluster-dev
12:15 Apeksha joined #gluster-dev
12:15 nbalacha joined #gluster-dev
12:23 * kkeithley thinks nigelb has a strange idea of what constitutes fun. (as in fun fact NetBSD has pkill)
12:24 nigelb kkeithley: :D
12:25 nigelb kkeithley: I've made interesting progress on the maintainer ack script.
12:25 nigelb It's gone in directions I didn't think it would go.
12:25 nigelb (I've only written the barest of logic)
12:25 kkeithley nigelb++ for the progress part
12:25 glusterbot kkeithley: nigelb's karma is now 21
12:25 nigelb And wrote a lot of tests
12:25 nigelb Catching all the odd cases.
12:25 rraja joined #gluster-dev
12:25 kkeithley sounds good
12:25 nigelb so, I've gone in this direction - we have the ability to define some paths as needing maintainer acks
12:26 nigelb If we have a change not in this folder, the script returns None (will vote 0)
12:27 nigelb If we have a change in paths where we want maintainer ack, someone defined as top-level architect or the maintainer needs to give +1 otherwise vote -1
12:27 kkeithley oh, it just registered....  maintainer ack script. hmm.  I guess I know what that does
12:27 nigelb Can you read python unit tests?
12:27 nigelb that's probably easier than me explaining.
12:27 kkeithley yes, I can read python. Kinda like I can read German.
12:28 suliba joined #gluster-dev
12:28 nigelb http://dpaste.com/2VFY9V8
12:29 nigelb the yml file to define paths and owners - http://dpaste.com/1KSZWXV
12:29 nigelb It's very rough. I've only written bare-minimum business logic.
12:29 kkeithley I think you can delete avati from that list.
12:30 nigelb I've coped a few entries from MAINTAINERS
12:31 kkeithley yes...  we should delete avati from MAINTAINERS too.
12:31 nigelb indeed.
12:32 * kkeithley wonders if something thinks he might come back some day
12:32 hagarth joined #gluster-dev
12:38 ndevos kkeithley: please say "zehn" for me out loud
12:39 * kkeithley looks around to see if ndevos has teleported to Massachusetts
12:39 ndevos nigelb: btw, I've added some comments in your etherpad
12:39 kkeithley okay, I said it.  Now if I click my heels will I be teleported to Berlin.
12:40 nigelb I noticed. I've incorporated some. I'll incorporate some at another point.
12:40 kkeithley madchen and bißchen are harder for me to say
12:40 nigelb kkeithley: It would be worth defining some rules for maintainership. If you're not active in the last X months, you'll lose your maintainer status+access
12:40 kkeithley define "active"
12:41 nigelb doing reviews/committing code
12:41 nigelb Might also be worth actually defining
12:41 nigelb Because I'd also say blog posts is "active"
12:41 ndevos kkeithley: its just that "zehn" sounds so cute when you say it :D
12:42 kkeithley me? cute? I don't think so.
12:42 kkeithley :P
12:43 ndevos heh, yes, exactly! nobody expects you to be cute, until they hear you speaking german
12:43 kkeithley lol
12:43 kkeithley meine Deutsch ist sehr gut
12:43 kkeithley Deutsche gesprechen
12:43 ndevos hmm, my arm/chromebook does not have automount in the kernel, seems I'm back to 'make menucofig' and compiling again :-/
12:44 kkeithley nicht wahr?
12:45 ndevos ja, dine Deutsch ist sehr gut!
12:45 ndevos *dein
12:46 ndevos ich muss aber lernen wie ich die Sonderzeichen tippen kann
12:46 post-factum nerozumim nemecky :(
12:47 ndevos mein US Tastatuerlayout hat keine Uemlaeute usw
12:47 kkeithley was beduite Sonderzeichen?
12:47 ndevos Zeichen mit Umlaut, und das S-Z
12:48 ndevos ist ja schwierig zu zeigen wenn es die Tasten nicht auf mein Tastatur gibt
12:48 misc wow, so I can actually understand german better than I believed
12:49 kkeithley Ich kann umlaute tippen  ëü ËÜ.
12:49 ndevos das hat nur eine weile gedauert :P
12:49 post-factum in russian ё is also the letter
12:49 kkeithley Meine Franzoche ist nicth so gut
12:49 post-factum but ё != ë
12:50 misc post-factum: ah ah, i guess no one abuse that for phising :)
12:50 ndevos oh, lol, und e mit Umlaut gibt es eigentlich gar nicht
12:50 post-factum misc: indeed, it seems you've never seen spam in russian
12:50 kkeithley Franzosishe
12:50 post-factum misc: in fact, it is widely used, and not only ё
12:50 kkeithley Franzosiche
12:50 ndevos auch ist ein Umlaut zwei Strichen, nicht Punkten, aber dass ist nur ein Detail
12:51 ndevos Franzoesisch?
12:51 ndevos oe = "o" mit Umlaut
12:51 kkeithley ja
12:51 misc post-factum: well, my antispam is efficient :)
12:51 post-factum misc: mine one too, but crap happens
12:52 ndevos und mein Franzoesisch ist auch nicht gut, dafuer gehe ich auch kaum nach Frankreich
12:52 kkeithley but compose+o+e in English is œ ligature
12:54 kkeithley what in Cyrillic looks like ë ?
12:54 post-factum it is pronounced like «yo»
12:55 post-factum in russian
12:55 kkeithley I guess it's the Cyrillic IE
12:55 post-factum ukrainian language does not have ё letter
12:56 kkeithley sorry, Cyrillic IO
12:56 post-factum yep
12:57 post-factum instead, ukrainian language have the cutest letter in the world — ї
12:57 post-factum *has
12:58 post-factum czech equiv is «ji». like in «jidlo» word
12:58 ndevos ... if only my screen/irssi/... supported utf-8 well enough
12:58 post-factum haha but it should ;)
12:59 kkeithley do you run it in utf-8 locale?
13:00 post-factum ndevos: do you see «АВЕКОРСТХ» to be almost the same as «ABEKOPCTX»?
13:01 post-factum those are common letters for both latin and cyrillic world
13:01 ndevos post-factum: my pop-up shows it just fine, but in irssi the 1st is just garbage
13:01 post-factum ndevos: oh, i missed "screen" word
13:01 post-factum i though, noone uses it in 2016
13:02 post-factum use tmux instead
13:02 ndevos well, its running on a RH hosted RHEL-6....
13:02 post-factum as if it is some form of excuse
13:02 kkeithley Looks fine in quassel.  Only К looks a little different than K
13:02 post-factum kkeithley: correct, bottom stroke is shifted a bit
13:03 post-factum and i missed «Н» — «H» pair
13:04 nixpanic on my CentOS-7 server with tmux+irssi thinks look good
13:04 post-factum and «І» — «I», «М» — «M»
13:04 post-factum probably, there is some fix for screen
13:06 ndevos I think there is, but it may not be in RHEL-6 :-/
13:07 post-factum does "screen -U" work for you?
13:07 ndevos I have "defutf8 on" in my .screenrc
13:07 post-factum i see some weird manuals mentioning one need to enable some form of recording from one of those iso-blablabla locales...
13:07 ndevos that should be the same, right?
13:07 kkeithley ♩♪♫ ♬
13:08 post-factum sing :)
13:08 post-factum ndevos: http://xkr47.outerspace.dyndns.o​rg/howtos/irssi-utf-8-guide.txt ?
13:08 post-factum «←→↑↓»
13:10 ndevos post-factum: thanks, but I think I've tried the -U switch before too - and I do not mind too much on missing out the non-ascii part
13:10 nixpanic and I can always check somewhere else
13:15 Apeksha joined #gluster-dev
13:32 jiffin1 joined #gluster-dev
13:36 aravindavk joined #gluster-dev
13:46 inevity joined #gluster-dev
13:47 hagarth joined #gluster-dev
14:28 dlambrig_ left #gluster-dev
14:29 shaunm joined #gluster-dev
14:42 msvbhat joined #gluster-dev
14:46 kkeithley hmmm.  merde.  The reason people can't build on Debian 8 (and Fedora 24 too fwiw) when they don't run autogen.sh first.....
14:47 kkeithley buildaux/config.{sub,guess}.dist
14:47 ndevos well, it should not be required to run that, the tarball should contain all
14:47 kkeithley that's true
14:48 kkeithley but beside the point
14:48 ndevos true, I think "can't build" is a little strong
14:49 kkeithley the tarball we ship, created by `make dist`, replaces config.{guess,sub} with the ones in build-aux.
14:49 ndevos its good to know that it happens on Fedora 24 too, that makes testing easier :)
14:50 ndevos oh, I thought we decided to not include those files?
14:50 * ndevos can be wrong, he doesnt remember the details that clearly
14:51 kkeithley I don't remember such a decision, but that is also beside the point.  We do seem to be shipping them regardless
14:51 ndevos ah, that was http://review.gluster.org/14503
14:52 ndevos so, those scripts are dummies... fedoras %configure macro replaces those
14:52 ndevos at least, it used to do that, maybe it changed?
14:52 kkeithley hmmm, okay
14:53 kkeithley don't know
14:54 ndevos ah, I'm really forgetfull, see https://github.com/gluster/glusterfs/blo​b/release-3.8/doc/release-notes/3.8.0.md
14:54 kkeithley anyway, the herefile in those is what's breaking the generated Makefiles
14:54 shyam joined #gluster-dev
14:55 kkeithley yeah, there's a lot to remember; I can't remember everything either.
14:55 kkeithley what am I looking for in those relnotes?
14:55 ndevos it's mentioned in the release notes that we expect config.guess and config.sub to be replaced by the distributions buid system
14:55 ndevos the 2nd paragraph or so
14:56 kkeithley is that were that change is introduced
14:56 kkeithley aha
14:56 ndevos but, if you went that far, you might have an idea how to decently return an error message instead of awkward build failures?
14:57 post-factum ndevos: http://review.gluster.org/#/c/15082/ == 15037+15100 for review
14:58 post-factum also i have some concerns about maintainers not backporting obvious and trivial things...
14:59 ndevos thanks post-factum, but 15100 is not merged yet :)
14:59 post-factum ndevos: i've cc'ed you because you did the review on it
15:00 post-factum ndevos: and 15100 is a fix for already merged 15037
15:00 kkeithley so one way to replace config.{sub,guess} is to run autogen.sh ;-)
15:00 ndevos post-factum: we're trying to only get backports for real bugs, and we should not spend too much time on trivial changes
15:00 post-factum ndevos: that is sad
15:01 ndevos post-factum: well, there is a reason we release new major versions, all non-important changes are included there
15:01 post-factum ndevos: real fix — 15068 — was not backported as well
15:01 post-factum i've submitted it as 15102
15:02 ndevos post-factum: we're putting backport criteria together in https://github.com/gluster/glusterdocs/pull/139
15:02 ndevos see https://github.com/kshlm/glusterdocs/blob/e7​29dd186df5938b2de2fc6fd5600be2736ee990/Contr​ibutors-Guide/GlusterFS-Release-process.md for a readable version
15:03 post-factum ndevos: ctrl+f "leak" shows nothing there :)
15:03 dlambrig joined #gluster-dev
15:03 post-factum i'm concerned, because memory leaking is the primary reason for me why i'm here
15:04 post-factum anyway, i'm going to pick such changes as long as i can
15:05 ndevos post-factum: yes, fixes like 15102 should get automatically backported by the developers and/or maintainers of the component
15:07 post-factum ok, gtg, have a nice weekend
15:07 shubhendu joined #gluster-dev
15:07 ndevos same to you post-factum!
15:09 jiffin joined #gluster-dev
15:10 ndevos I think this test-case is stable, running from this morning, and I ow have: Test #1113 passed
15:10 ndevos that means, no failures yet :)
15:11 kkeithley ndevos: do you remember the _real_ rationale behind http://review.gluster.org/#/c/14503/ ? Was it because we were shipping jurassic era config.{sub,guess} files?
15:11 wushudoin joined #gluster-dev
15:13 ndevos kkeithley: yes, I think so, and we do not want to keep updating them whenever - several distributions ship patched versions to enable new arches
15:13 kkeithley Even though this was just merged in May of this year I can't figure out, can't remember, enough context to decide if this was a good decision or not
15:13 ndevos I think the decision is good, just the execution is lacking a little
15:14 ndevos it would be best to get a real error message when ./configure is run
15:15 ndevos if that message explains that config.{guess,sub} needs to be replaced by the distributions version, or one could run ./autogen.sh
15:17 shyam joined #gluster-dev
15:19 kkeithley right now I'm kinda feeling like what we did was/is a hack.  If only because, as one example, gcc ships 'real' config.{sub,guess}, and if it's good enough for gcc, maybe it should be good enough for us.
15:20 kkeithley especially now that we would be shipping recent versions of those files.
15:21 ndevos hey shyam!
15:21 ndevos shyam: I'm reworking gfapi/upcall to be more safe, it'll look like http://termbin.com/j95r
15:21 shyam ndevos: here (but in a meeting, expect slow responses)
15:22 ndevos shyam: np, thats just a preview of the change I'm sending soon
15:23 ndevos kkeithley: if we ship them, we need some way to watch out for updates to them - there have been CVE's against those at one point too, I thought
15:23 ndevos kkeithley: its one of the reasons Fedora/RHEL replaces them with their own copies
15:24 ndevos kkeithley: glfs_free() and safer memory usage in gfapi with http://termbin.com/j95r - I still have to work out a backward compatible solution
15:24 kkeithley what's on there now are from RHEL
15:25 ndevos from the redhat-rpm-config RPM, or automake/config?
15:25 ndevos (well, not sure if it was redhat-rpm-config, maybe an other similar package?)
15:27 kkeithley automake-1.15-1.el6.noarch  appears to be the source of config.{guess,sub}
15:27 ndevos and redhat-rpm-config contains the copies that %configure uses
15:27 ndevos maybe they have the same contents?
15:28 aravindavk joined #gluster-dev
15:28 kkeithley nope
15:28 ndevos [Other Topic] where did glfs_ipc() come from? it seems to use our dict_t structure that we do not expose...
15:29 kkeithley IIRC that came from jdarcy
15:30 ndevos sometimes libgfapi feels like a bastard amoung all of the components we have :-/
15:30 kkeithley and yes, I objected to that
15:30 ndevos its also labelled as version 4.0.0, that seems missfitting too
15:30 ndevos oh, or did I revert that for 3.8?
15:32 kkeithley did it make it into 3.8?
15:32 ndevos no, I do not think so
15:32 ndevos I did revert several 4.0.0 things, maybe I caught this too, or it wasnt merged at the time
15:33 ndevos but it surely is something we need to tell aravindavk and pranith about - they need to revert the 4.0 changes as well
15:34 ndevos kkeithley: oh, YOU revered it with bug 1345977
15:34 glusterbot Bug https://bugzilla.redhat.com:​443/show_bug.cgi?id=1345977 unspecified, unspecified, ---, kkeithle, CLOSED CURRENTRELEASE, api: revert glfs_ipc_xd intended for 4.0
15:34 kkeithley mkay
15:34 * ndevos clones that bug for 3.9 now
15:43 ndevos kkeithley: any early objections to http://termbin.com/j95r ?
15:45 kkeithley meh.  looks to me like a release would ship the /usr/share/automake-1.15/config.{sub,guess} if we let it.
15:46 kkeithley And at a guess if we made a src.rpm that /usr/lib/rpm/redhat/brp-redhat might replace those with /usr/lib/rpm/redhat/config.{sub,guess}
15:46 ndevos yes, correct
15:47 kkeithley quick look, no objections
15:47 ndevos and then we get complaints from Debian people that config.{sub,guess} is not current enough for some of the architectures they support
15:47 kkeithley are those really all meant to be for "public" consumption?
15:48 ndevos you mean things like glfs_upcall_get_reason() etc? yes, those are what Ganesha needs to use for cache-invalidation
15:48 ndevos of course I still need to add documentation about those functions...
15:49 kkeithley sure, but I was thinking more along the lines of "do we want anyone/everyone using them?"   The distinction between public and private, as it applies to versioned symbols, is really non-existent.
15:50 ndevos yes, any application that wants to use upcalls should use those public functions
15:50 kkeithley private is kind of a warning flag that most people shouldn't use them
15:50 kkeithley there's no enforcement
15:50 ndevos we just made them rather unstable with the first incarnation, and this should make it much safer, and we can change the internals if we like
15:51 ndevos sure, but why should we suggest to not use those functions?
15:58 ndevos and on that note, glfs_ipc_xd should be marked private, nothing outside of the glusterfs sources can use dict_t
16:05 pranithk1 joined #gluster-dev
16:27 kkeithley if those functions are declared in glfs.h or glfs-handles.h then they should be public. The existing priv(ate) functions are not declared in those headers, but are used by gluster.
16:28 kkeithley I'm just asking the question, do they need to be public?  Apparently they do.
16:28 ndevos yes, they are in glfs-handles.h
16:28 ndevos at least Ganesha needs to use them, otherwise we cant do cache-invalidation
16:28 kkeithley and if they don't need to be public, then the decls should not be in those headers.  But (at the risk of repeating) apparently they do
16:29 kkeithley They could be private for Ganesha only.
16:30 ndevos that does not really make sense to me, we *could*, but what benefit would it have?
16:30 ndevos the current (broken) callback/upcall stuff is public as well
16:31 ndevos this just replaces that
16:31 kkeithley Are they intended for everyone to use? Or are they really only for Ganesha?
16:32 ndevos Ganesha is the first that wants to do cache invalidation, I do not see exclusive rights for Ganesha for that functionality
16:32 kkeithley This is just like declaring a function in a C file static or not.  Only at a different level.
16:32 kkeithley okay, then they should be public.
16:33 ndevos good, glad thats settled then :)
16:34 kkeithley now, about reverting the config.{sub,guess} change....
16:38 ndevos I prefer not to revert it, but give a clear error when running ./configure and tell the user to replace the files, or run ./autogen.sh
16:38 ndevos pretty sure that was intended with the change, but the error is probably hidden in config.log
16:39 shyam1 joined #gluster-dev
16:41 kkeithley the error is not exactly hidden. ;P
16:41 kkeithley How to fix it is
16:42 kkeithley s/is/may be/
16:44 ndevos also, its 18:40 here already, and I Need to be on a squash court in 20 minutes
16:44 kkeithley anyway, I go back to my gcc arguement. If it's good enough for gcc, it should be good enough for us.  gcc's config.{sub,guess} appear to be slightly newer than what's in automake-1.15.
16:45 kkeithley have fun, good luck
16:45 ndevos we would still need to update is with whatever version $distribution needs...
16:45 kkeithley ttyl
16:45 ndevos I prefer not to include those files, and give a warning when the user did not replace them
16:46 ndevos the warning is in config.log now (I think), and that should be printed on the console when running ./configure
16:47 ndevos cya!
17:14 spalai joined #gluster-dev
17:43 sanoj joined #gluster-dev
18:12 dlambrig left #gluster-dev
18:16 uebera|| joined #gluster-dev
18:27 penguinRaider joined #gluster-dev
18:29 ramky joined #gluster-dev
18:59 skoduri joined #gluster-dev
19:00 jiffin joined #gluster-dev
19:21 jiffin1 joined #gluster-dev
21:08 shyam joined #gluster-dev
22:02 rastar_ joined #gluster-dev
22:08 xavih_ joined #gluster-dev
22:27 tdasilva joined #gluster-dev

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