Bug 50174 – Get rid of prefix header dependency for WebKit2 build system

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKit2

[Laszlo Gombos]

WebKit(1) does not need "prefix header" support from the build system (see config.h), but WebKit2 has this dependency. This bug is to track building WebKit2 without "prefix header" support from the compiler.

The topic has been discussed in the webkit-dev mailing list - https://lists.webkit.org/pipermail/webkit-dev/2010-July/013427.html.

RVCT 2.2 with the current Symbian (and likely other embedded) build system does not seems to support "prefix header". Builds with icecc would also take advantage of this work.

[Balazs Kelemen]

I agree that a config.h style solution would be more straightforward.
The last feedback I got from the Apple guys was that they do not want
another config.h for WebKit2. However, distcc and icecc are happy with 
the prefix header at least with gcc (only precompiled header is a 
problem). Does RVCT 2.2 really luck the  -include option or something 
equivalent?

[Andras Becsi]

(In reply to comment #1)
> Does RVCT 2.2 really lack the  -include option or something equivalent?

RVCT 4.0 has a -include option but drops a warning:
Warning: C3910W: Old syntax, please use '-I'.

So I guess previous versions also supported -include, but it might behave like -I, which prepends a directory to the include path.

As far as I know RVCT has a --pch option since 2.0, and at least 4.0 has a
--use_pch option, which should behave similar to gcc's -include.

[Balazs Kelemen]

Gabor(s), do you know how to simulate -include with rvct?

[Laszlo Gombos]

RVCT (including 2.2) does have a compiler option for prefix headers - it is called "--preinclude". There is more on this here: http://www.keil.com/support/man/docs/armccref/armccref_CHDIIAEI.htm

However there is a bug in the 2.2 version of RVCT that it only takes the last instance of the preinclude header into account (which typically points to RVCT2_2.h). I have not found a way to convince RVCT to take our own preinclude header with the existing Qt/Symbian toolchain.

As noted RVCT is not the only environment where this is an issue.

[Siddharth Mathur]

If you happen to compile for Symbian OS, then a provisional solution with RVCT 2.2 is to add the following line at the end of your $$EPOCROOT/epoc32/include/rvct/rvct.h

#ifdef BUILDING_WEBKIT
#include "..\..\..\webkit\WebKit2\WebKit2Prefix.h"
#endif

[Greg]

Created attachment 77351 [details]
Adding config.h to every WebKit2 cpp file.

[Early Warning System Bot]

Attachment 77351 [details] did not build on qt:
Build output: http://queues.webkit.org/results/7257124

[Build Bot]

Attachment 77351 [details] did not build on win:
Build output: http://queues.webkit.org/results/7235150

[David Levin]

Comment on attachment 77351 [details]
Adding config.h to every WebKit2 cpp file.

View in context: https://bugs.webkit.org/attachment.cgi?id=77351&action=review

r- because this breaks some builds. 

Also, does this change have consensus with the folks working on WebKit2? This seems like it should be resolved on webkit-dev (rather than a bug/patch).

> Tools/Scripts/webkitpy/style/checker.py:161
> +    # Certain directories have other idiosyncracies.

Nit: "other" would no longer apply here.

[Greg]

Thanks for the review David, I'm looking into the build failures.

The topic has been raised on webkit-dev mailing list - 
https://lists.webkit.org/pipermail/webkit-dev/2010-November/015162.html . 

No notable objections has been received so far.

[Greg]

Created attachment 79772 [details]
Adding config.h to every WebKit2 cpp file.

[Greg]

Created attachment 79797 [details]
Adding config.h to every WebKit2 cpp file.

[WebKit Review Bot]

Attachment 79797 [details] did not pass style-queue:

Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files', u'Source/WebCore/ChangeLog', u'Source/WebCor..." exit_code: 1

Source/WebKit2/config.h:117:  "WebCore/config.h" already included at Source/WebKit2/config.h:28  [build/include] [4]
Total errors found: 1 in 361 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Balazs Kelemen]

Looks absolutely good to me but we should convince the Apple folks about
the need of this change.

[Laszlo Gombos]

This looks good to me as well. I plan to r+ it and land it on Thursday, unless someone object.

[Laszlo Gombos]

Comment on attachment 79797 [details]
Adding config.h to every WebKit2 cpp file.

r=me. cq- as the patch needs to be re-baselined.

[Laszlo Gombos]

Created attachment 80402 [details]
rebaselined patch for cq to commit

[WebKit Commit Bot]

Comment on attachment 80402 [details]
rebaselined patch for cq to commit

Rejecting attachment 80402 [details] from commit-queue.

Failed to run "['./Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '--bot-id=eseidel-cq-sl', 'ap..." exit_code: 2

Last 500 characters of output:
h
|+++ Source/WebKit2/config.h
--------------------------
No file to patch.  Skipping patch.
1 out of 1 hunk ignored
patching file Source/WebKit2/win/WebKit2Common.vsprops
Hunk #1 FAILED at 6.
1 out of 1 hunk FAILED -- saving rejects to file Source/WebKit2/win/WebKit2Common.vsprops.rej
patching file Tools/ChangeLog
Hunk #1 succeeded at 1 with fuzz 3.
patching file Tools/Scripts/webkitpy/style/checker.py

Failed to run "[u'/Projects/CommitQueue/Tools/Scripts/svn-apply', u'--force']" exit_code: 1

Full output: http://queues.webkit.org/results/7542395

[Laszlo Gombos]

Created attachment 80404 [details]
try again, this time patch generated from svn

[WebKit Commit Bot]

Comment on attachment 80404 [details]
try again, this time patch generated from svn

Rejecting attachment 80404 [details] from commit-queue.

Failed to run "['./Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '--bot-id=eseidel-cq-sl', 'ap..." exit_code: 2

Last 500 characters of output:
cess/qt/WebProcessQt.cpp
patching file Source/WebKit2/WebProcess/win/WebProcessMainWin.cpp
patching file Source/WebKit2/WebProcess/win/WebProcessWin.cpp
patching file Source/WebKit2/win/WebKit2Common.vsprops
Hunk #1 FAILED at 6.
1 out of 1 hunk FAILED -- saving rejects to file Source/WebKit2/win/WebKit2Common.vsprops.rej
patching file Tools/ChangeLog
patching file Tools/Scripts/webkitpy/style/checker.py

Failed to run "[u'/Projects/CommitQueue/Tools/Scripts/svn-apply', u'--force']" exit_code: 1

Full output: http://queues.webkit.org/results/7525396

[Laszlo Gombos]

Created attachment 80415 [details]
rebaseline WebKit2Common.vsprops

[WebKit Commit Bot]

Comment on attachment 80415 [details]
rebaseline WebKit2Common.vsprops

Rejecting attachment 80415 [details] from commit-queue.

Failed to run "['./Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '--bot-id=eseidel-cq-sf', 'ap..." exit_code: 2

Last 500 characters of output:
cess/qt/WebProcessQt.cpp
patching file Source/WebKit2/WebProcess/win/WebProcessMainWin.cpp
patching file Source/WebKit2/WebProcess/win/WebProcessWin.cpp
patching file Source/WebKit2/win/WebKit2Common.vsprops
Hunk #1 FAILED at 6.
1 out of 1 hunk FAILED -- saving rejects to file Source/WebKit2/win/WebKit2Common.vsprops.rej
patching file Tools/ChangeLog
patching file Tools/Scripts/webkitpy/style/checker.py

Failed to run "[u'/Projects/CommitQueue/Tools/Scripts/svn-apply', u'--force']" exit_code: 1

Full output: http://queues.webkit.org/results/7546380

[WebKit Review Bot]

http://trac.webkit.org/changeset/76916 might have broken SnowLeopard Intel Release (Tests)

[Laszlo Gombos]

Comment on attachment 80415 [details]
rebaseline WebKit2Common.vsprops

Landed manually as http://trac.webkit.org/changeset/76916.