799458 - Error: package error or possible missing or unnecessary file: bin/components/libnkgnomevfs.so (package-manifest, 172).

Status: RESOLVED FIXED

(SeaMonkey :: Build Config, defect)

Description

[Tony Mechelynck [:tonymec]]

Attachment: mozconfig

In L64 SeaMonkey-trunk ownbuild id=20121008144954 c-c:876f51d9458f m-c:876f51d9458f, during "make package":
Error: package error or possible missing or unnecessary file: bin/components/libnkgnomevfs.so (package-manifest, 172).

I'll attach the mozconfig & build log (not quote them since probably of marginal importance).

Comment 1

[Tony Mechelynck [:tonymec]]

Attachment: build log (bz2 compressed)

NB. Both zip and gz compression formats are half again as long.

Comment 2

[Tony Mechelynck [:tonymec]]

Oops: moz-c revision 9738e5a0190a

Comment 3

[Frank Wein [:mcsmurf]]

Probably "caused" by this check-in: http://hg.mozilla.org/mozilla-central/rev/157f3efdee37 we need to sync this check-in from mozilla-central configure.in to comm-central.

Comment 4

[Frank Wein [:mcsmurf]]

Attachment: Patch

Comment 5

[Frank Wein [:mcsmurf]]

Comment on attachment 669775 [details] [diff] [review]
Patch

Test run on try server: https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=3cc5a4c4d541

Comment 6

[Tony Mechelynck [:tonymec]]

(In reply to Frank Wein [:mcsmurf] from comment #5)
> Comment on attachment 669775 [details] [diff] [review]
> Patch
> 
> Test run on try server:
> https://tbpl.mozilla.org/?tree=Thunderbird-Try&rev=3cc5a4c4d541

That looks like a Thunderbird build. I'll try to compile SeaMonkey here at home with the patch applied.

Don't forget to run qrefresh -U -m 'bug 799458 - package error or possible missing or unnecessary file: bin/components/libnkgnomevfs.so' (or similar) before you set checkin-needed (if ever). (Assuming you use the mq extension to Mercurial)

Comment 7

[Justin Wood (:Callek)]

Comment on attachment 669775 [details] [diff] [review]
Patch

I'll leave this pending tony's local build.

Comment 8

[Tony Mechelynck [:tonymec]]

Attachment: build log with attachment 669775

Build proceeds to the end, but I don't have enough disk space to unpack the tar.bz2 in /usr/local (didn't try removing <objdir>/mozilla/dist/seamonkey/). Neither have I enough disk quota to upload the .tar.bz2 where you guys could lay hands on it.

Trying to run it straight as "<objdir>/mozilla/dist/seamonkey/seamonkey -P default" gives me a startup crash (bp-31b7e5e2-bd10-48f2-8e89-d7a0e2121010, no symbols that Socorro can see) at startup, after checking the extensions for compatibility with version 2.16a1.

Comment 9

[Tony Mechelynck [:tonymec]]

If I change my mozconfig for --enable-install-strip, that should spare some disk space; OTOH if I don't I might perhaps get a usable crash trace with gdb somehow.

Comment 10

[Tony Mechelynck [:tonymec]]

Attachment: console log leading to the crash

Not sure if relevant.

Comment 11

[Tony Mechelynck [:tonymec]]

Attachment: end of the log under gdb

This was obtained under gdb; it was a SIGSEGV, and a stack trace (with symbols, except for shared system libraries) is included. It immediately follows attachment 669945 [details].

At the time of the crash, the frame of the ChatZilla window was already visible; not yet the browser & mailer.

And I still don't know if it's relevant.

Comment 12

[Frank Wein [:mcsmurf]]

The crash does not look related to my change. If a crash happens because of this patch, I would rather expect to see something with shell service or Gnome in the stack trace. Maybe you can try again with a clean build?

Comment 13

[Tony Mechelynck [:tonymec]]

(In reply to Frank Wein [:mcsmurf] from comment #12)
> The crash does not look related to my change. If a crash happens because of
> this patch, I would rather expect to see something with shell service or
> Gnome in the stack trace. Maybe you can try again with a clean build?

huhuh. I'll see when I can squeeze it in. If there were a pause of some kind just before the libxul link I would know when to unload everything else.

Comment 14

[Tony Mechelynck [:tonymec]]

Mozilla/5.0 (X11; Linux x86_64; rv:19.0) Gecko/19.0 Firefox/19.0 SeaMonkey/2.16a1 ID:20121011064956 c-c:8a86f48e53ea m-c:2fae8bd461da

(mcsmurf's try build) starts normally in my default profile (checks extensions for 2.16a1 compat, then opens cZ, M&N, browser and opens my 67 tabs including this one).

Comment 15

[Frank Wein [:mcsmurf]]

BTW: Tony meant this build http://ftp.mozilla.org/pub/mozilla.org/thunderbird/try-builds/[e-mail]-8a86f48e53ea/ I created it on the Thunderbird try server (you need to modify the mail/config/mozconfigs/* files by replacing --enable-application=mail with =suite and remove "--skip-chatzilla --skip-inspector --skip-venkman" from build/client.py-args).

Comment 16

[Frank Wein [:mcsmurf]]

Pushed to comm-central: https://hg.mozilla.org/comm-central/rev/4f0f19f411b4

Comment 17

[Tony Mechelynck [:tonymec]]

Linux-x86_64 builds are of course still burning because of bug 798921; yet this is already one fewer hurdle to jump.

Comment 18

[Tony Mechelynck [:tonymec]]

Since then, all hourlies and nightlies have become green again. Most tests are orange or even red or violet but I suppose this is to be expected.

Frank, I think that the bug was just forgotten and should have been RESOLVED FIXED long ago. If I goofed, feel free to REOPEN (with a comment saying what must still be done in this bug).

Comment 19

[Iacopo Benesperi [:iacchi]]

I see that target milestone is v. 2.16: does this mean that I won't be able to compile SeaMonkey 2.15 on Linux? Because I've just downloaded the sources for SeaMonkey 2.15 and following the same building procedure I've used for over an year, compile goes ok but using make inside the installer folder gets me the error subject of this bug. Is this supposed to be?

Comment 20

[Frank Wein [:mcsmurf]]

Hrm, looks like this patch should have also landed on SeaMonkey 2.15. Not sure why we did not, although it's certainly not a critical issue that we forgot about this (the official builds build with gnomevfs as this has been enabled in the official mozconfig because of other reasons). I suggest you download the patch (Attachment 669775 [details] [diff]) and apply the patch in the top source folder with "patch -p1<bug_799458".

Comment 21

[Frank Wein [:mcsmurf]]

Though this might not be so good for Linux distributors that build their own SeaMonkey :/
BTW: If you build without the patch, it will fail to build/include gnomevfs in the resulting .tar.gz file. By default SeaMonkey uses GIO though instead of gnomevfs and GIO is built into libxul as far as I see this. So by default this error is harmless.

Comment 22

[Frank Wein [:mcsmurf]]

Oh and when you manually enable gnomevfs with --enable-gnomevfs/--disable-gio in .mozconfig, it will work anyway. So that error message is harmless (but might be a bit confusing).

Comment 23

[Iacopo Benesperi [:iacchi]]

With the patch applied it worked fine but it asked me, at first run, if I wanted to use it as the default program for... Non a big problem, but I thought it worth telling.
Shouldn't be better to release a 2.15.1 or something with this patch applied? Not being able to build on linux is not a minor issue imho.

Comment 24

[Frank Wein [:mcsmurf]]

From what I know a packaging error is not fatal (as in, the packaging process stops when such a error occurs). And yes, SeaMonkey can now be set as default client/browser on Linux, too (see release notes http://www.seamonkey-project.org/releases/seamonkey2.15/changes)

Comment 25

[Frank Wein [:mcsmurf]]

No need to track this anymore.