955009 - Ship libpurple as an add-on and get Instantbird nightlies building off comm-central

Status: RESOLVED FIXED

(Chat Core :: General, enhancement)

Description

[Patrick Cloke [:clokep]]

*** Original post on bio 1579 at 2012-07-11 18:36:00 UTC ***

We don't depend on libpurple anymore, GTalk, Facebook, IRC and (kind of) XMPP can be used without it. Eventually we'd like the build system to be changed to build libpurple as an add-on (or at least to disable it). This should allow an MPL 2 only build of Instantbird. (And would allow Thunderbird to take an add-on implementing libpurple.)

Comment 1

[Florian Quèze [:florian]]

Attachment: WIP1

*** Original post on bio 1579 as attmnt 1832 at 2012-08-23 14:25:00 UTC was without comment, so any subsequent comment numbers will be shifted ***

Comment 2

[Florian Quèze [:florian]]

Attachment: WIP2

*** Original post on bio 1579 as attmnt 1833 at 2012-08-23 17:52:00 UTC ***

This builds, but the libpurple file isn't loaded before libpurplexpcom and can't be loaded automatically because it's not in the folder of the main executable.

Comment 3

[Florian Quèze [:florian]]

Comment on attachment 8353591 [details] [diff] [review]
WIP1

*** Original change on bio 1579 attmnt 1832 at 2012-08-23 17:52:14 UTC was without comment, so any subsequent comment numbers will be shifted ***

Comment 4

[Florian Quèze [:florian]]

Attachment: WIP3

*** Original post on bio 1579 as attmnt 1859 at 2012-08-27 17:56:00 UTC ***

This works on Mac.
With an amount of hacks that makes me wonder if this approach is worth pursuing.

Comment 5

[Florian Quèze [:florian]]

Comment on attachment 8353592 [details] [diff] [review]
WIP2

*** Original change on bio 1579 attmnt 1833 at 2012-08-27 17:56:02 UTC was without comment, so any subsequent comment numbers will be shifted ***

Comment 6

[Florian Quèze [:florian]]

Attachment: WIP3.1

*** Original post on bio 1579 as attmnt 1864 at 2012-08-28 13:42:00 UTC ***

Same as WIP3 but with an additionnal mozilla patch that I forgot to hg add before making the previous diff.

I think I'm going to hold off on this, as I think the downsides of this patch (have to rerun configure to update changed makefiles in purple/ as for some reason the makefile update script fails when entering the directory; complicated and confusing build order of the various folders; various hacks in the makefiles that will be a pain to maintain and will break each time our build system gets out of sync with the mozilla/ one, ...) far outweight the (limited) benefits.

Comment 7

[Florian Quèze [:florian]]

Comment on attachment 8353618 [details] [diff] [review]
WIP3

*** Original change on bio 1579 attmnt 1859 at 2012-08-28 13:42:48 UTC was without comment, so any subsequent comment numbers will be shifted ***

Comment 8

[Florian Quèze [:florian]]

*** Original post on bio 1579 at 2012-08-28 13:49:40 UTC ***

Pasting here what I just said on IRC:
I think a way better solution would be to put the whole purple/ folder in a separate repository, and have the folder layout for it match what's done for venkman and DOM i, and checkout that repository from client.py.

In our current situation having 2 separates code repositories for the same application would be painful, so I think that approach would be worth it only if we get the greenlight for integrating instantbird/ in comm-central.

Comment 9

[Patrick Cloke [:clokep]]

(In reply to Florian Quèze [:florian] [:flo] from comment #8)
> In our current situation having 2 separates code repositories for the same
> application would be painful, so I think that approach would be worth it
> only if we get the greenlight for integrating instantbird/ in comm-central.

Now that this was done in bug 956609, I started looking at this a bit. I started with the same procedure as bug 956609: copying the purple/ directory from Instantbird (I assume we'd want to eventually hg convert the history) and applying patches to get things building. I'm going to attach my current WIPs.

Comment 10

[Patrick Cloke [:clokep]]

Attachment: 3 - Update Makefile.in and moz.build for glib and xml2.

Before trying these patches you'll need to copy the current purple source (which is too big to attach to Bugzilla):
cp -r ../instantbird/purple mozilla/extensions && cd mozilla/extensions/purple && hg init && hg add . && hg commit -m "Importing initial purple/ folder."

Comment 11

[Patrick Cloke [:clokep]]

Attachment: 2 - Add JAR_MANIFESTS to moz.build.

Comment 12

[Patrick Cloke [:clokep]]

Attachment: 4 - Update Makefile.in and moz.build for libpurple

This patch is definitely a WIP and it does NOT work yet, but it does some of the basic porting (CSRCS to SOURCES, etc.)

Comment 13

[Florian Quèze [:florian]]

Attachment: purple bundle

Bundle resulting of hg converting the purple/ history (also renaming to get rid of the "purple" folder and have the content of the current purple's folder directly at the top of the repository).

Comment 14

[Patrick Cloke [:clokep]]

Attachment: 2.2 - Add JAR_MANIFESTS to moz.build

Comment 15

[Patrick Cloke [:clokep]]

Attachment: 3.2 - Update Makefile.in and moz.build for glib and xml2

This gets glib and xml2 to build, looks like there's a few commented out lines of code which we can remove (hence f? instead of r?).

Comment 16

[Patrick Cloke [:clokep]]

Attachment: 4.2 - Update Makefile.in and moz.build for libpurple

Comment 17

[Patrick Cloke [:clokep]]

Attachment: 5 - Convert prpl-rules.mk to prpl.py for use in moz.build

Currently nothing is built as a static prpl since configure doesn't work.

Comment 18

[Patrick Cloke [:clokep]]

Attachment: 6 - Update Makefile.in and moz.build for purplexpcom

Comment 19

[Patrick Cloke [:clokep]]

Attachment: 7 - Fix up purplexpcom's memory reporters

This patch was provided by Florian and I've reviewed it. It seems to build, although I haven't tested it in a running Instantbird yet.

Comment 20

[Patrick Cloke [:clokep]]

Attachment: 8 - Port |Bug 956493 - Replace PRUnichar with char16_t|

This ports Bug 956493, I had to modify the script it used already, but it's pretty straightforward.

Comment 21

[Patrick Cloke [:clokep]]

Attachment: 9 - Port |Bug 910989 - nsTHashtable::Init should go away|

This ports bug 910989 for purplexpcom.

Comment 22

[Patrick Cloke [:clokep]]

Attachment: 10 - Copy version.h from current Instantbird

This copies version.h from my current Instantbird build. I think we could still dynamically generate this if we played with moz.build/Makefile.in a bit (and with help from #build). I'd probably prefer that solution, but it doesn't really upset me to check this in.

Comment 23

[Patrick Cloke [:clokep]]

Attachment: A - Run purple's configure from Instantbird

This will run configure from mozilla/extension/purple (and is a patch for c-c, not for the purple/ repo as everything else is).

Comment 24

[Patrick Cloke [:clokep]]

Attachment: 11 WIP - Add configure

This just grabs the old configure rules from Instantbird and puts them into a patch for purple.

Things left to do:
- Figure out how to actually get configure to run for purple/.
- Ensure static purples are built.
- Ensure everything compiles properly on all platforms.

I'd appreciate some help looking at the configure stuff as I'm unsure of where to go next with getting configure to run properly.

Comment 25

[Florian Quèze [:florian]]

Comment on attachment 8369228 [details] [diff] [review]
3.2 - Update Makefile.in and moz.build for glib and xml2

Review of attachment 8369228 [details] [diff] [review]:
-----------------------------------------------------------------

Seems good.

::: libraries/glib/moz.build
@@ +175,5 @@
> +FINAL_LIBRARY = 'purplexpcom'
> +
> +LIBRARY_NAME = 'glib'
> +
> +#FORCE_SHARED_LIB = True

Remove this line.

::: libraries/xml2/moz.build
@@ +60,5 @@
> +FINAL_LIBRARY = 'purplexpcom'
> +
> +LIBRARY_NAME = 'xml2'
> +
> +#FORCE_SHARED_LIB = True

and this one.

Comment 26

[Florian Quèze [:florian]]

Comment on attachment 8369229 [details] [diff] [review]
4.2 - Update Makefile.in and moz.build for libpurple

Review of attachment 8369229 [details] [diff] [review]:
-----------------------------------------------------------------

::: libpurple/Makefile.in
@@ +17,5 @@
>  ifeq (,$(filter WINNT Darwin,$(OS_ARCH)))
>  LOCAL_INCLUDES	= $(GLIB_CFLAGS) $(LIBXML2_CFLAGS)
>  else
>  LOCAL_INCLUDES  = -I$(DIST)/include/glib
> +#SHARED_LIBRARY_LIBS += $(foreach lib,glib xml2,../libraries/$(lib)/$(LIB_PREFIX)$(lib).$(LIB_SUFFIX))

Remove this.

::: libpurple/moz.build
@@ +141,5 @@
> +FINAL_LIBRARY = 'purplexpcom'
> +
> +LIBRARY_NAME = 'purple'
> +
> +#FORCE_STATIC_LIB = True

Remove this.

Comment 27

[Florian Quèze [:florian]]

Comment on attachment 8369230 [details] [diff] [review]
5 - Convert prpl-rules.mk to prpl.py for use in moz.build

Review of attachment 8369230 [details] [diff] [review]:
-----------------------------------------------------------------

::: libpurple/protocols/prpl.py
@@ +1,5 @@
> +# This Source Code Form is subject to the terms of the Mozilla Public
> +# License, v. 2.0. If a copy of the MPL was not distributed with this
> +# file, You can obtain one at http://mozilla.org/MPL/2.0/.
> +
> +if CONFIG['STATIC_PRPLS'] and protocol in CONFIG['STATIC_PRPLS']:

This test checks that you are in the 'statically linked prpl' case...

@@ +2,5 @@
> +# License, v. 2.0. If a copy of the MPL was not distributed with this
> +# file, You can obtain one at http://mozilla.org/MPL/2.0/.
> +
> +if CONFIG['STATIC_PRPLS'] and protocol in CONFIG['STATIC_PRPLS']:
> +    SOURCES += ['xpcomModule.cpp']

but this is a file you need to make the prpl a dynamically loaded xpcom component.

@@ +4,5 @@
> +
> +if CONFIG['STATIC_PRPLS'] and protocol in CONFIG['STATIC_PRPLS']:
> +    SOURCES += ['xpcomModule.cpp']
> +
> +    FORCE_STATIC_LIB = True

I think this isn't needed any more, and we use FINAL_LIBRARY = 'purplexpcom' instead?

@@ +9,5 @@
> +
> +LIBRARY_NAME = protocol
> +
> +if CONFIG['PURPLEXPCOM']:
> +    FORCE_SHARED_LIB = True

Here is where you should add the xpcomModule.cpp file I guess.

@@ +11,5 @@
> +
> +if CONFIG['PURPLEXPCOM']:
> +    FORCE_SHARED_LIB = True
> +
> +IS_COMPONENT = True

This is also something you need only when building a dynamically loaded xpcom component. Statically linked prpls aren't xpcom components.

Comment 28

[Florian Quèze [:florian]]

Comment on attachment 8369231 [details] [diff] [review]
6 - Update Makefile.in and moz.build for purplexpcom

Review of attachment 8369231 [details] [diff] [review]:
-----------------------------------------------------------------

::: purplexpcom/src/Makefile.in
@@ +15,3 @@
>  #NO_DIST_INSTALL	= 1
>  
> +#SHARED_LIBRARY_LIBS += ../../libpurple/$(LIB_PREFIX)purple.$(LIB_SUFFIX)

Remove this please. And the 2 commented out lines above can go too.

::: purplexpcom/src/moz.build
@@ +41,5 @@
> +EXTRA_COMPONENTS += ['prpl.manifest']
> +
> +IS_COMPONENT = True
> +
> +#FINAL_LIBRARY = 'purplexpcom'

Remove this.

Comment 29

[Florian Quèze [:florian]]

Comment on attachment 8369235 [details] [diff] [review]
10 - Copy version.h from current Instantbird

Review of attachment 8369235 [details] [diff] [review]:
-----------------------------------------------------------------

I'm OK with doing this, but I think we need to handle generating this file in the libpurple update script. If you don't do it now, I think you should at least add an |echo| line reminding the user of the script that this needs to be done by hand for now.

Comment 30

[Florian Quèze [:florian]]

Comment on attachment 8369237 [details] [diff] [review]
A - Run purple's configure from Instantbird

Review of attachment 8369237 [details] [diff] [review]:
-----------------------------------------------------------------

This would break building c-c with --enable-application=im without checking out libpurple.

I think we will want to add the purple extension through a line in our default mozconfigs, rather than in confvars.sh

Comment 31

[Patrick Cloke [:clokep]]

Attachment: 3.3 - Update Makefile.in and moz.build for glib and xml2

(In reply to Florian Quèze [:florian] [:flo] from comment #25)
> Comment on attachment 8369228 [details] [diff] [review]
> 3.2 - Update Makefile.in and moz.build for glib and xml2
> 
> Review of attachment 8369228 [details] [diff] [review]:

I removed the commented out lines.

Comment 32

[Patrick Cloke [:clokep]]

Attachment: 4.3 - Update Makefile.in and moz.build for libpurple

(In reply to Florian Quèze [:florian] [:flo] from comment #26)
> Comment on attachment 8369229 [details] [diff] [review]
> 4.2 - Update Makefile.in and moz.build for libpurple
> 
> Review of attachment 8369229 [details] [diff] [review]:
> -----------------------------------------------------------------
I removed the commented out lines.

Comment 33

[Patrick Cloke [:clokep]]

Attachment: 6.2 - Update Makefile.in and moz.build for purplexpcom

(In reply to Florian Quèze [:florian] [:flo] from comment #28)
> Comment on attachment 8369231 [details] [diff] [review]
> 6 - Update Makefile.in and moz.build for purplexpcom
> 
> Review of attachment 8369231 [details] [diff] [review]:
> -----------------------------------------------------------------

I also removed the commented out lines from here.

Comment 34

[Patrick Cloke [:clokep]]

Attachment: 10.2 - Copy version.h from current Instantbird

(In reply to Florian Quèze [:florian] [:flo] from comment #29)
> Comment on attachment 8369235 [details] [diff] [review]
> 10 - Copy version.h from current Instantbird
> 
> Review of attachment 8369235 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> I'm OK with doing this, but I think we need to handle generating this file
> in the libpurple update script. If you don't do it now, I think you should
> at least add an |echo| line reminding the user of the script that this needs
> to be done by hand for now.

I added a statement at the end of upgrade-libpurple.sh that says to update version.h, but does not say how to do it. This can probably be scripted easily, I'll add it when I do bug 964828.

Comment 35

[Patrick Cloke [:clokep]]

Attachment: A.2 - Build purple extension in mozconfigs

(In reply to Florian Quèze [:florian] [:flo] from comment #30)
> Comment on attachment 8369237 [details] [diff] [review]
> A - Run purple's configure from Instantbird
> 
> Review of attachment 8369237 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> This would break building c-c with --enable-application=im without checking
> out libpurple.
> 
> I think we will want to add the purple extension through a line in our
> default mozconfigs, rather than in confvars.sh
OK, so I added --enable-extension=purple in im/configs/mozconfigs/*. Is this OK?

Comment 36

[Florian Quèze [:florian]]

Comment on attachment 8370414 [details] [diff] [review]
A.2 - Build purple extension in mozconfigs

Review of attachment 8370414 [details] [diff] [review]:
-----------------------------------------------------------------

--enable-extensions, not --enable-extension

Comment 37

[Patrick Cloke [:clokep]]

Attachment: A.3 - Build purple extension in mozconfigs

(In reply to Florian Quèze [:florian] [:flo] from comment #36)
> Comment on attachment 8370414 [details] [diff] [review]
> A.2 - Build purple extension in mozconfigs
> 
> Review of attachment 8370414 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> --enable-extensions, not --enable-extension

Sorry, I copied this out of my .mozconfig wrong.

Comment 38

[Patrick Cloke [:clokep]]

Attachment: 5.2 - Convert prpl-rules.mk to prpl.py for use in moz.build

This is an updated version of prpl.py that fixes the logic in the review comments.

Things to do at this point are:
- configure
- packaging
- Figure out how to get dynamic prpls to compile after libpurple

Comment 39

[Florian Quèze [:florian]]

Comment on attachment 8369232 [details] [diff] [review]
7 - Fix up purplexpcom's memory reporters

># HG changeset patch
># User Florian Quèze <[e-mail]>
># Date 1391034663 18000
>#      Wed Jan 29 17:31:03 2014 -0500
># Node ID e7aadccc1aa9b9eb4be02cc0dc69d3900ced33de
># Parent  be70c165e9d591ba4c480a5b628ec9d7f65e624f
>Bug 955009 -  Ship libpurple as an add-on / allow it to be disabled in configure, Rewrite purplexpcom memory reporters for interface changes, r=clokep.

hg import didn't like this attachment:
abort: decoding near 'Florian Qu?ze <flori': 'utf8' codec can't decode byte 0xe8 in position 10: invalid continuation byte!

Comment 40

[Patrick Cloke [:clokep]]

Attachment: B - Call purple configure from im

With help from aleth and Florian, not requesting review on this yet.

Comment 41

[Patrick Cloke [:clokep]]

Attachment: 11.2 - Add configure.in to purple

This seems to run, but I don't see any output configure file. Just uploading this here to get us all on the same page.

Comment 42

[Patrick Cloke [:clokep]]

Attachment: 4.4 - Update Makefile.in and moz.build for libpurple

This uses GENERATED_SOURCES instead of SOURCES for static_proto_init.c.

Comment 43

[Florian Quèze [:florian]]

Comment on attachment 8379191 [details] [diff] [review]
11.2 - Add configure.in to purple

Review of attachment 8379191 [details] [diff] [review]:
-----------------------------------------------------------------

::: configure.in
@@ +4,5 @@
> +dnl License, v. 2.0. If a copy of the MPL was not distributed with this
> +dnl file, You can obtain one at http://mozilla.org/MPL/2.0/.
> +
> +dnl Get version of libpurple from the version files.
> +PURPLE_MAJOR_VERSION=`cat $topsrcdir/extensions/purple/config/version.txt |cut -d '.' -f 1`

I think I would vaguely prefer:
purpledir=$topsrcdir/extensions/purple
PURPLE_MAJOR_VERSION=`cat $purpledir/config/version.txt |cut -d '.' -f 1`

But this is fine either way :-).

Comment 44

[Florian Quèze [:florian]]

Attachment: B.2 - Call purple's subconfigure from im/configure.in, and packaging changes

Comment 45

[Florian Quèze [:florian]]

Attachment: 12 - Fix building prpls

Comment 46

[Patrick Cloke [:clokep]]

Comment on attachment 8379364 [details] [diff] [review]
B.2 - Call purple's subconfigure from im/configure.in, and packaging changes

Review of attachment 8379364 [details] [diff] [review]:
-----------------------------------------------------------------

This looks fine, I'm testing this on Windows now. Do we want some way to automatically download the purple repository (once we figure out where that lives...?)

Comment 47

[Patrick Cloke [:clokep]]

Attachment: 11.3 - Add configure.in to purple

Carrying this review forward, it adds the purpledir variable, as requested.

Comment 48

[Patrick Cloke [:clokep]]

Comment on attachment 8379366 [details] [diff] [review]
12 - Fix building prpls

Review of attachment 8379366 [details] [diff] [review]:
-----------------------------------------------------------------

::: purplexpcom/src/moz.build
@@ +5,5 @@
>  
>  XPCSHELL_TESTS_MANIFESTS += ['test/xpcshell.ini']
>  
> +prpls = ['jabber', 'gg', 'oscar', 'msn', 'myspace', 'novell',
> +         'qq', 'sametime', 'simple', 'yahoo', 'netsoul']

I dislike having this full list of prpls in here twice. (It's also in libpurple/moz.build.) I think we should go w/ this for now if it works, but it'd be nice if there was a way to simplify this.

Comment 49

[Patrick Cloke [:clokep]]

(In reply to Patrick Cloke [:clokep] from comment #48)
> Comment on attachment 8379366 [details] [diff] [review]
> 12 - Fix building prpls
> 
> Review of attachment 8379366 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> ::: purplexpcom/src/moz.build
> @@ +5,5 @@
> >  
> >  XPCSHELL_TESTS_MANIFESTS += ['test/xpcshell.ini']
> >  
> > +prpls = ['jabber', 'gg', 'oscar', 'msn', 'myspace', 'novell',
> > +         'qq', 'sametime', 'simple', 'yahoo', 'netsoul']
> 
> I dislike having this full list of prpls in here twice. (It's also in
> libpurple/moz.build.) I think we should go w/ this for now if it works, but
> it'd be nice if there was a way to simplify this.

Sorry, this comment makes no sense, please disregard it!

Comment 50

[Florian Quèze [:florian]]

(In reply to Patrick Cloke [:clokep] from comment #46)

> Do we want some way to automatically download the purple repository

Eventually/ideally client.py should handle it. For now I was thinking that we would just do it with buildbot.

Comment 51

[Patrick Cloke [:clokep]]

Comment on attachment 8353622 [details] [diff] [review]
WIP3.1

I successfully got this to build and run on Windows!

Comment 52

[Florian Quèze [:florian]]

Comment on attachment 8369234 [details] [diff] [review]
9 - Port |Bug 910989 - nsTHashtable::Init should go away|

Review of attachment 8369234 [details] [diff] [review]:
-----------------------------------------------------------------

In bug 910989 comment 3 roc says:
> There are a number of places that lazily initialize nsTHashtables, and
> sometimes the initialization status (via IsInitialized) as some kind of
> state flag. [...] I'm wrapping the hashtable in
> nsAutoPtr and heap-allocating it at the point where it would have been
> lazily initialized.

Let's follow this.

In purpleGetText.h, change:
    nsDataHashtable<nsCStringHashKey, nsTArray<nsCString> *> table;
to:
    nsAutoPtr<nsDataHashtable<nsCStringHashKey, nsTArray<nsCString> *> > table;

(You may need to include nsAutoPtr.h)

::: purplexpcom/src/purpleGetText.cpp
@@ -286,5 @@
> -      // If we have an entry but the hashtable isn't initialized,
> -      // it means that the translation bundle for this package is
> -      // missing. The error value will make the translation code
> -      // return early.
> -      if (!mStringCache[i].table.IsInitialized())

Make this a simple null check:
if (!mStringCache[i].table)

@@ -326,5 @@
>      return -1;
>    }
>  
>    // The translation file exists, init the hashtable and return the new index
> -  pack->table.Init();

Make this create the hashtable:
pack->table = new nsDataHashtable<nsCStringHashKey, nsTArray<nsCString> *>();

@@ -512,5 @@
>      os->RemoveObserver(sInstance, NS_CHROME_FLUSH_TOPIC);
>  
>    LOG(("purpleGetText::unInit\n"));
>    for (PRUint32 i = 0; i < sInstance->mStringCache.Length(); ++i)
> -    if (sInstance->mStringCache[i].table.IsInitialized())

if (sInstance->mStringCache[i].table)

@@ -513,5 @@
>  
>    LOG(("purpleGetText::unInit\n"));
>    for (PRUint32 i = 0; i < sInstance->mStringCache.Length(); ++i)
> -    if (sInstance->mStringCache[i].table.IsInitialized())
> -      sInstance->mStringCache[i].table.Enumerate(deleteEnumerator, nullptr);

table.Enumerate becomes table->Enumerate

(also replace . by -> in all the table.Put and table.Get throughout the file)

Comment 53

[Patrick Cloke [:clokep]]

Attachment: 9.2 - Port |Bug 910989 - nsTHashtable::Init should go away|

This makes the requested changes, this compiles and ran.

Comment 54

[aleth [:aleth]]

Attachment: WIP: Add purple repo to client.py

This adds the purple repo to client.py in the "obvious" way. It defaults to skipping it (i.e. one would have to use "python client.py --include-purple checkout").

Hasn't been properly tested as the repo doesn't exist yet.

Comment 55

[aleth [:aleth]]

Attachment: WIP: Add purple repo to client.py v2

Fixes a help string. It may be premature to request feedback considering it hasn't been fully tested, but is this the right direction to go?

Comment 56

[Patrick Cloke [:clokep]]

Florian created and pushed to a repo: http://hg.mozilla.org/users/florian_queze.net/purple

Comment 57

[Florian Quèze [:florian]]

https://hg.mozilla.org/comm-central/rev/f33bbb41b11c
https://hg.mozilla.org/comm-central/rev/116c2a4f5b6e

Comment 58

[aleth [:aleth]]

Attachment: Add purple repo to client.py

Updated with the correct repo path now it exists. Seems to work as expected.

Comment 59

[Florian Quèze [:florian]]

mozconfig fixup: https://hg.mozilla.org/comm-central/rev/659fa1b366aa

Comment 60

[Joshua Cranmer [:jcranmer]]

Comment on attachment 8383575 [details] [diff] [review]
Add purple repo to client.py

Review of attachment 8383575 [details] [diff] [review]:
-----------------------------------------------------------------

So the current plan for ccrework will be that we drop client.py altogether from the normal build invocation process. If you're planning on using this to set up the buildbots, I'd advise you to not do so. If you want it for ease of developer-side machines, I'd advise asking gps et al about adding it to m-c client.py instead.

Comment 61

[Florian Quèze [:florian]]

(In reply to Joshua Cranmer [:jcranmer] from comment #60)
> If you're planning on using this
> to set up the buildbots, I'd advise you to not do so.

We were more or less hoping to use this to simplify our buildbot configs, but I've actually started reconfiguring our buildbots for the changes made in this bug, and for now I added buildbot steps to deal with this additional repository.

> If you want it for
> ease of developer-side machines, I'd advise asking gps et al about adding it
> to m-c client.py instead.

This is more the real goal.

Comment 62

[Florian Quèze [:florian]]

The xpcshell of the purple/ folder aren't running, in the log we have this warning:
Included file '/Users/buildbot/buildslave/macosx/obj-instantbird/i386/mozilla/_tests/xpcshell/purple/purplexpcom/src/test/xpcshell.ini' does not exist

Comment 63

[aleth [:aleth]]

(In reply to Florian Quèze [:florian] [:flo] from comment #61)
> > If you want it for
> > ease of developer-side machines, I'd advise asking gps et al about adding it
> > to m-c client.py instead.
> 
> This is more the real goal.

Requesting info on whether adding an equivalent of this patch to the m-c client.py would be acceptable.

Comment 64

[Florian Quèze [:florian]]

More follow-ups:

https://hg.mozilla.org/comm-central/rev/8b9bd8b734ce - remove obsolete optimization flags from the Windows buildbot mozconfigs. The -GL flag was breaking the build of libvpx.

https://hg.mozilla.org/comm-central/rev/8cd4fcbbbb5e - fix linux buildbot mozconfigs for compatibility with our CentOS5 builders. This disables pulseaudio. I also enabled gnomevfs which apparently still works as a gio replacement (discovered this while looking at the SeaMonkey mozconfigs).

https://hg.mozilla.org/comm-central/rev/f8112386067e - fix buildsymbols. It was broken on Windows because of a path issue. And it wasn't finding the libpurple source repository.

Comment 65

[Patrick Cloke [:clokep]]

http://hg.mozilla.org/users/florian_queze.net/purple/rev/6b77268d791c - Fix building of purpleCoreService on Linux

Please file other issues in new bugs.