Bug 53766 – [Qt] Compilation error on OpenBSD

• Component: Platform
• Status: RESOLVED FIXED

[Olivier Goffart]

src/3rdparty/webkit/WebCore/websockets/WebSocketHandshake.cpp:289: error: 'strnstr' was not declared in this scope

That function is apparently not protbale. According to the man page:
"Since the strnstr() function is a FreeBSD specific API, it should only be used when portability is not a concern."

[Patrick R. Gansterer]

Please add a OS(OPENBSD) (or whatever) at http://trac.webkit.org/browser/trunk/Source/JavaScriptCore/wtf/StringExtras.h?rev=74855#L103

[Benjamin Poulain]

(In reply to comment #1)
> Please add a OS(OPENBSD) (or whatever) at http://trac.webkit.org/browser/trunk/Source/JavaScriptCore/wtf/StringExtras.h?rev=74855#L103

I would be happy to but I can't make a patch I do not test. Patrick, could you make the patch for OpenBSD and test the build?

Here is some info about making patches: https://trac.webkit.org/wiki/QtWebKitContrib
Ping me on IRC if you have any question.

[Patrick R. Gansterer]

(In reply to comment #2)
> I would be happy to but I can't make a patch I do not test.
I have the same "problem" of no OpenBSD installation.

> Patrick, could you make the patch for OpenBSD and test the build?
I'm be willing to create a patch, but Oliver needs to test it first.

Mabye my last comment should be read as:
"Oliver, can you try to add OS(OPENBSD) to StringExtras.h and tell us if that fixes the problem"
Sorry for the unclear comment.

[Olivier Goffart]

I confirm that adding || OS(OPENBSD) in StringExtras.h works.

Then webkit compile and i am able to run the Qt demo browser.

[Patrick R. Gansterer]

Created attachment 82021 [details]
Patch

[Benjamin Poulain]

Comment on attachment 82021 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=82021&action=review

> Source/JavaScriptCore/wtf/Platform.h:729
> +#if !defined(HAVE_STRNSTR)
> +#if !COMPILER(MSVC) && !COMPILER(RVCT) && !OS(WINDOWS) && !OS(LINUX) && !OS(SOLARIS) && !OS(OPENBSD)

Shouldn't that be the other way around:
#if OS(MAC) || OS(FREEBSD)
?

[Patrick R. Gansterer]

Created attachment 82034 [details]
Alternative Patch (with positiv listing)

View in context: https://bugs.webkit.org/attachment.cgi?id=82021&action=review

>> Source/JavaScriptCore/wtf/Platform.h:729
>> +#if !COMPILER(MSVC) && !COMPILER(RVCT) && !OS(WINDOWS) && !OS(LINUX) && !OS(SOLARIS) && !OS(OPENBSD)
> 
> Shouldn't that be the other way around:
> #if OS(MAC) || OS(FREEBSD)
> ?

Maybe yes, maybe no. ;-)
I don't know if it's defined at AIX, HAIKU, NETBSD, QNX, but changing this to positiv listing is clearly better.

It seams that at least HAIKU needs this change too, so it's ok to take the risk of breaking the build on some "uncommon" platforms IMHO.

[Andreas Kling]

Comment on attachment 82034 [details]
Alternative Patch (with positiv listing)

r=me

[Andreas Kling]

(In reply to comment #7)
> It seams that at least HAIKU needs this change too, so it's ok to take the risk of breaking the build on some "uncommon" platforms IMHO.

Judging by the qt-haiku port (I had it handy..), Haiku doesn't have strnstr(). http://gitorious.org/webkit/qt-haiku-webkit/commit/b964b5fb44144d48a8bce266db856d3317d12a0f?diffmode=sidebyside

[WebKit Commit Bot]

Comment on attachment 82034 [details]
Alternative Patch (with positiv listing)

Clearing flags on attachment: 82034

Committed r78278: <http://trac.webkit.org/changeset/78278>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[WebKit Review Bot]

http://trac.webkit.org/changeset/78278 might have broken GTK Linux 32-bit Release
The following tests are not passing:
http/tests/misc/acid2-pixel.html
http/tests/misc/acid2.html