1562504 - comm-central build error - A jar.mn exists but it is not referenced in the moz.build file. Please define JAR_MANIFESTS.

Status: RESOLVED INVALID

(Thunderbird :: Build Config, defect)

Description

[ISHIKAWA, Chiaki]

Attachment: JAR_MANIFESTS.patch

I needed to modify a couple of moz.build files to proceed with build after a new source tree was fetched last Friday or so.

A jar.mn exists but it is not referenced in the moz.build file. Please define JAR_MANIFESTS.

A patch is attached.

I am not sure if that is the optimum patch: I wonder if I needed to change other files instead of these leaf moz.build files.

TIA
PS: I have other errors due to very strict compiler flag I use to make sure my patches will compile on try server and am thus in the middle of applying the local patches to avoid compiler warning/errors.
At least, the changes to moz.build files were necessary to configure and run compiler.

Comment 1

[Magnus Melin [:mkmelin]]

I'm confused. There are no jar.mn files in those dirs.

Comment 2

[ISHIKAWA, Chiaki]

(In reply to Magnus Melin [:mkmelin] from comment #1)

I'm confused. There are no jar.mn files in those dirs.

Oh, then, it could be HG's "hg pull -u" conflict issue, then.
Ever since I enabled auto clang-format feature using "mach" (I am not sure if I HAVE enabled it locally), I have enough problems of merge issues. Maybe I set it wrong. I have actually fetched a source tree afresh from the ground up a couple of times :-(
Maybe this is another update failure possibly.

env LANG=C ls -l /NREF-COMM-CENTRAL/mozilla/comm/mail/components/devtools/ lists the following.
total 28
-rw-r--r-- 1 ishikawa ishikawa 2811 Jun 27 13:43 devtools-loader.js
-rw-r--r-- 1 ishikawa ishikawa 429 Mar 12 06:44 devtools-loader.manifest
-rw-r--r-- 1 ishikawa ishikawa 646 Jun 12 00:32 jar.mn
-rw-r--r-- 1 ishikawa ishikawa 388 Jun 27 23:16 moz.build
-rw-r--r-- 1 ishikawa ishikawa 2685 Jun 27 13:43 tb-root-actor.js
-rw-r--r-- 1 ishikawa ishikawa 514 Jun 12 00:32 toolbox-process-overlay.xul
-rw-r--r-- 1 ishikawa ishikawa 622 Jun 12 00:32 webconsole-overlay.xul

env LANG=C ls -l /NREF-COMM-CENTRAL/mozilla/comm/mail/extensions/smime &
total 12
drwxr-xr-x 2 ishikawa ishikawa 4096 Jun 27 13:44 content
-rw-r--r-- 1 ishikawa ishikawa 1931 Mar 12 06:44 jar.mn
-rw-r--r-- 1 ishikawa ishikawa 256 Mar 12 06:44 moz.build

I am afraid I have to re-fetch the source tree from the ground up again and start over... :-(

TIA

Comment 3

[Magnus Melin [:mkmelin]]

There is no jar.mn file in mailnews/extensions/smime/, and there is none in mail/components/devtools/
There is one in comm/mail/extensions/smime, but that's now what you have in your patch.

I wonder what this says about the file:
hg log mail/components/devtools/jar.mn

Comment 4

[ISHIKAWA, Chiaki]

(In reply to Magnus Melin [:mkmelin] from comment #3)
...

I wonder what this says about the file:
hg log mail/components/devtools/jar.mn

Here it is. I had to add env LANG=C to make the date appear in non-Japanese locale format.

ishikawa@ip030:/NREF-COMM-CENTRAL/mozilla/comm$ env LANG=C hg log mail/components/devtools/jar.mn

changeset: 20700:f85659ef3cf0
user: Philipp Kewisch <[e-mail]>
date: Sat Sep 24 13:24:37 2016 +0200
summary: Bug 1279834 - Integrate devtools client UI into Thunderbird. r=mkmelin

changeset: 19443:4da79677dea3
user: Philipp Kewisch <[e-mail]>
date: Fri Jun 10 01:02:24 2016 +0200
summary: Fix bug 1243550 - Bring Browser Console to Thunderbird. r=mkmelin

changeset: 13119:9791b8bc1e64
user: Philipp Kewisch <[e-mail]>
date: Sat Sep 28 14:48:10 2013 +0200
summary: Bug 884805 - Figure out if/how to best package the debugger as an extension. r=mconley

Strange that the last update was in 2016. Definitely a corrupt repository?

TIA

Comment 5

[Magnus Melin [:mkmelin]]

Try "hg purge"
Also hg pull -u && hg up default

Comment 6

[ISHIKAWA, Chiaki]

(In reply to Magnus Melin [:mkmelin] from comment #5)

Try "hg purge"
Also hg pull -u && hg up default

"hg purge" seems to have helped.

CAUTION: to anyone reading this exchange, "hg purge" removes ALL FILES that are present (config files, shell scripts) that are present in the source tree and NOT in the central repository. BEWARE (!)
I had mozconfig and a few shell scripts in the source tree, which have been wiped out, but to my luck, I saved them under a different directory.
So one's development habit (whether storing config files and shell scripts in the source tree where they are used, whether the private files are managed using the same VCS as the original source tree, etc. affects the effectiveness/peril of running "hg purge".

hg pull -u && hg up default
also seems to have helped.
But there were some strange interaction of |hg|.
I fetched |comm| sub tree uner |mozilla| to start afresh.
Now I forgot exactly what this some strange interaction (very minor) was.

After downloading |comm| subtree afresh before attempting to apply my local patches by re-creating them from the stored old-tree(s) and
merge tool, I thought of updating the devtools by |mach bootstrap| and
upon doing that, enabled clang-format and js-format extension: I thought this was a good idea
for the source tree which sees more of the automatic formatting all the time.

Disaster. These extensions don't work well with HG MQ usage patterns.
My newly created patches were trashed.

Bug 1564445
FATAL mach bootstrap: after accepting the change for clang-format and js-format in ~.hgrc, |hg qnew -Ue| now invokes editor four (4) times for editing commit message (-e) for MQ "hg qnew" each time and messes patch queue completely.

I will be back after a few days once I manage to compile the source tree again.

(PS: My local setup adds additional C++ warning/error flags to avoid
seeing compiler failure on the tryserver once my patches compile locally.
Many errors caused by the stricter flags are incorrect # of arguments of macros, incorrect arg vs format strings, sign vs. unsigned comparison, etc.
Anyway, I have to correct these before I can compile locally successfully, and patches to take care of these are no longer applied after source tree reformatting and I have to bring them up-to-date before I can work on really interesting bug fixes and enhancement patches which I have created locally...)

Comment 7

[ISHIKAWA, Chiaki]

Because of the above issue Bug 1564445, I had to refetch C-C source tree a couple of times anew and finally could compile with minimum set of patches to shut up compiler errors caused by very strict compiler flags for local build and testing.

Since the issue is no longer seen, I think the bug was a fluke caused by a corrupt mercurial tree.
So I am closing this bug as INVALID.

Thank you for the tips to put the tree in good shape.
The bug 1564445 really caused a serious disruption, though. :-(