[Webkit-unassigned] [Bug 31273] Added macros for USERINCLUDE paths within symbian blocks

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=31273





--- Comment #23 from Norbert Leser <norbert.leser at nokia.com>  2009-11-16 06:10:00 PST ---
(In reply to comment #22)
> (In reply to comment #21)
> 
> I haven't tried the patch yet so don't know if it broke anything.
> Headers are moved about in MCL releases and I would assume this could
> potentially break anyways.
> 
> > I agree with you that it would be preferable to apply this change to all
> > localheader include cases. I am not sure what you mean with using "userinclude"
> 
> MMP_RULES += "USERINCLUDE ."
> 
> or change each include not under epoc32\include to be user includes.
> 
> Would be the only 2 alternatives I can live with :)

Wouldn't your first option still require some of the webkit code to be changed
to define includes relative to ".", such as #include "profiler/profiler.h"?
That is what I wanted to avoid (I believe there was a request with this patch
earlier this year which was either rejected or reversed).

The 2nd option seems to be very invasive, if I understand that right (would
that require to special case for SYMBIAN almost all include paths in
JavaScripCore.pri?

> > 
> > The patch for these particular include paths on targets beyond 5.0 due to
> > changes in the (abld) tools chain. The first SYSTEMINCLUDE always becomes
> > /epoc32/include, regardless of declarations in mmp files. That is, we must use
> > USERINCLUDE to guarantee inclusion before that.
> 
> This was a bug in SBSv2 and also on abld and has been fixed from SBSv2 already.
> It was added to fix some silly test case without thinking...
> 
> Toolchain is bugged, but we are trying to fix it instead of trying to work
> around it.

I am with you, to fix the toolchain, but we are facing the problem of build
breaks right now and I have no idea when the fixed toolchain will be available.
Besides, it is not a specific but a structural issue that the symbian toolchain
uses USERINCLUDE paths instead of honoring the "localinclude.h" property. I
haven't seen any attempt to fix that, and we have to work around that, somehow,
in the least intrusive way (I still think, if we proof that these 3 USERINCLUDE
declarations of this patch don't break other targets, it seems to be a useful
workaround for the current situation).

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.