Workaround scons and xcode not understanding #include MESSAGES_INTERNAL_FILE.

Workaround scons and xcode not understanding #include MESSAGES_INTERNAL_FILE.

BUG=7828

[Craig Schlenter, 2009-02-20 05:53:19]

This is a bit of a hack and it may compile things
unnecessarily but I think it's better than nothing.

Reviewers please. Thank you.

[tony, 2009-02-27 21:24:14]

http://codereview.chromium.org/20529/diff/1003/1004
File chrome/common/ipc_message_macros.h (right):

http://codereview.chromium.org/20529/diff/1003/1004#newcode49
Line 49: // Trick scons into seeing the possible real dependencies since it
doesn't
This seems to work for xcode too.

http://codereview.chromium.org/20529/diff/1003/1004#newcode54
Line 54: #include "chrome/common/render_messages_internal.h"
Can you also add chrome/renderer/dev_tools_messages_internal.h and
chrome/test/automation/automation_messages_internal.h?

[Craig Schlenter, 2009-02-28 07:04:15]

On 2009/02/27 21:24:14, tony wrote:
[snip]
> This seems to work for xcode too.

I have updated the comment and the patch description to mention xcode now.

> Can you also add chrome/renderer/dev_tools_messages_internal.h and
> chrome/test/automation/automation_messages_internal.h?

Done.

Do you know what effect this has on visual studio btw.? - I expect it might end
up compiling a few things unnecessarily like scons and xcode will now. I'm
surprised that windows can apparently figure out the dependencies given that
there are multiple possible values of MESSAGES_INTERNAL_FILE ...

Thank you!

[tony, 2009-03-02 21:40:56]

I tested in Visual Studio.  It gets this case right.  I built everything,
touched dev_tools_messages_internal.h, then built again. Only
dev_tools_client.cc and dev_tools_agent.cc recompiled.  I assume VS runs through
the preprocessor before finding dependencies which is why it got it right in the
first place.

This LGTM.

John or Mark, do you have a strong opposition to this?

On 2009/02/28 07:04:15, Craig Schlenter wrote:
> On 2009/02/27 21:24:14, tony wrote:
> [snip]
> > This seems to work for xcode too.
> 
> I have updated the comment and the patch description to mention xcode now.
> 
> > Can you also add chrome/renderer/dev_tools_messages_internal.h and
> > chrome/test/automation/automation_messages_internal.h?
> 
> Done.
> 
> Do you know what effect this has on visual studio btw.? - I expect it might
end
> up compiling a few things unnecessarily like scons and xcode will now. I'm
> surprised that windows can apparently figure out the dependencies given that
> there are multiple possible values of MESSAGES_INTERNAL_FILE ...
> 
> Thank you!

[jam, 2009-03-02 21:43:52]

I don't have strong opposition, but I'm still curious why the scons method that
we discusses earlier isn't suitable?

http://codereview.chromium.org/20529/diff/4001/2002
File chrome/common/ipc_message_macros.h (right):

http://codereview.chromium.org/20529/diff/4001/2002#newcode52
Line 52: #include "chrome/common/ipc_sync_message_unittest.h"
why is this needed?

[tony, 2009-03-02 22:30:08]

On 2009/03/02 21:43:52, John Abd-El-Malek wrote:
> I don't have strong opposition, but I'm still curious why the scons method
that
> we discusses earlier isn't suitable?

I explained it on the bug:
http://code.google.com/p/chromium/issues/detail?id=7828&can=3&colspec=ID%20St...


> http://codereview.chromium.org/20529/diff/4001/2002
> File chrome/common/ipc_message_macros.h (right):
> 
> http://codereview.chromium.org/20529/diff/4001/2002#newcode52
> Line 52: #include "chrome/common/ipc_sync_message_unittest.h"
> why is this needed?

Because that file also has IPC messages in it (used in the
ipc_sync_channel_unittest.cc).

[tony, 2009-03-03 01:44:51]

Committed in r10772 from http://codereview.chromium.org/27374.  Thanks for the
patch!