OpenBSD/FreeBSD GYP changes (most of the remaining ones)

OpenBSD/FreeBSD GYP changes (most of the remaining ones)



Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=38079

[Peter Valchev, 2010-02-03 23:28:31]

I renamed the 'linux' files to 'posix' in printing/ and sdch/ but some of the
other ones like skia are going to be much more time consuming and I'd rather not
rename right now unless you feel very strongly (in some of the cases there were
already 'freebsd' checks to the linux cases and I just added 'openbsd' :)

sdch/bsd/config.h is a copy of the linux file, but I decided to add it just in
case they ever diverge (and for hygiene)

GYP is actually fighting me at the moment, the Makefiles generated do not seem
to respect the CPPFLAGS (so my build doesn't get very far), something must have
changed in the last two months that I can't yet figure out...

[Evan Martin, 2010-02-03 23:52:04]

Sorry to back-seat drive, but do you think it would make more sense to name the
printing files "foo_cairo.cc"?

http://codereview.chromium.org/565043/diff/1/23
File ipc/ipc.gyp (right):

http://codereview.chromium.org/565043/diff/1/23#newcode66
ipc/ipc.gyp:66: ['OS=="linux"', {
Good catch.

[Peter Valchev, 2010-02-04 00:17:22]

On 2010/02/03 23:52:04, Evan Martin wrote:
> Sorry to back-seat drive, but do you think it would make more sense to name
the
> printing files "foo_cairo.cc"?

cairo makes more sense, sure. I renamed them again, let's see if "svn rename"
twice in a row works or if I'll have to start it from scratch...

[Evan Martin, 2010-02-04 00:37:26]

looks like you maybe missed one rename:
  printing/pdf_ps_metafile_linux_unittest.cc

note that the trybots don't work right with renames, but you should be at least
able to use the trybots to verify you didn't break mac/win.  if you don't have a
linux box handy i can test this for you on mine.

http://codereview.chromium.org/565043/diff/1007/1011
File printing/pdf_ps_metafile_cairo.h (right):

http://codereview.chromium.org/565043/diff/1007/1011#newcode1
printing/pdf_ps_metafile_cairo.h:1: cairo// Copyright (c) 2009 The Chromium
Authors. All rights reserved.
typo

http://codereview.chromium.org/565043/diff/1007/1014
File printing/printing.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1014#newcode67
printing/printing.gyp:67: ['OS!="linux" and OS!="freebsd" and OS!="openbsd"',
{'sources/': [['exclude', '_cairo\\.cc$']]}],
80 cols

http://codereview.chromium.org/565043/diff/1007/1023
File sdch/bsd/config.h (right):

http://codereview.chromium.org/565043/diff/1007/1023#newcode5
sdch/bsd/config.h:5: /* #undef HAVE_DLFCN_H */
Is it really true that BSD has none of these functions?

http://codereview.chromium.org/565043/diff/1007/1029
File third_party/ffmpeg/ffmpeg.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1029#newcode10
third_party/ffmpeg/ffmpeg.gyp:10: ['OS!="linux" and OS!="freebsd" and
OS!="openbsd"', {'sources/': [['exclude', '/linux/']]}],
80 cols

http://codereview.chromium.org/565043/diff/1007/1029#newcode50
third_party/ffmpeg/ffmpeg.gyp:50: ['(OS!="linux" and OS!="freebsd" and
OS!="openbsd" and OS!="mac") or use_system_ffmpeg!=0', {
80 cols

http://codereview.chromium.org/565043/diff/1007/1028
File third_party/libxml/libxml.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1028#newcode8
third_party/libxml/libxml.gyp:8: ['OS=="linux" or OS=="freebsd" or
OS=="openbsd"', {'os_include': 'linux'}],
80

http://codereview.chromium.org/565043/diff/1007/1027
File third_party/libxslt/libxslt.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1027#newcode8
third_party/libxslt/libxslt.gyp:8: ['OS=="linux" or OS=="freebsd" or
OS=="openbsd"', {'os_include': 'linux'}],
80

[Peter Valchev, 2010-02-04 01:46:45]

Renamed the missed file too and uploaded new changeset... I did get the previous
renaming of linux->posix to work with the trybots but I had to have the right
sequence of rename, THEN modify before creating the CL... so I just reverted
everything, and renamed them in the right sequence, let's see if the trybots
like me...

http://codereview.chromium.org/565043/diff/1007/1011
File printing/pdf_ps_metafile_cairo.h (right):

http://codereview.chromium.org/565043/diff/1007/1011#newcode1
printing/pdf_ps_metafile_cairo.h:1: cairo// Copyright (c) 2009 The Chromium
Authors. All rights reserved.
On 2010/02/04 00:37:26, Evan Martin wrote:
> typo

oops thanks

http://codereview.chromium.org/565043/diff/1007/1014
File printing/printing.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1014#newcode67
printing/printing.gyp:67: ['OS!="linux" and OS!="freebsd" and OS!="openbsd"',
{'sources/': [['exclude', '_cairo\\.cc$']]}],
On 2010/02/04 00:37:26, Evan Martin wrote:
> 80 cols

Done.

http://codereview.chromium.org/565043/diff/1007/1023
File sdch/bsd/config.h (right):

http://codereview.chromium.org/565043/diff/1007/1023#newcode5
sdch/bsd/config.h:5: /* #undef HAVE_DLFCN_H */
On 2010/02/04 00:37:26, Evan Martin wrote:
> Is it really true that BSD has none of these functions?

I copied the linux file. :)

http://codereview.chromium.org/565043/diff/1007/1029
File third_party/ffmpeg/ffmpeg.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1029#newcode10
third_party/ffmpeg/ffmpeg.gyp:10: ['OS!="linux" and OS!="freebsd" and
OS!="openbsd"', {'sources/': [['exclude', '/linux/']]}],
On 2010/02/04 00:37:26, Evan Martin wrote:
> 80 cols

Done.

http://codereview.chromium.org/565043/diff/1007/1029#newcode50
third_party/ffmpeg/ffmpeg.gyp:50: ['(OS!="linux" and OS!="freebsd" and
OS!="openbsd" and OS!="mac") or use_system_ffmpeg!=0', {
On 2010/02/04 00:37:26, Evan Martin wrote:
> 80 cols

embarrassingly i'm not sure how to wrap this line and keep the syntax correct.
\ & newline doesn't seem to do it and neither does just newline?

http://codereview.chromium.org/565043/diff/1007/1028
File third_party/libxml/libxml.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1028#newcode8
third_party/libxml/libxml.gyp:8: ['OS=="linux" or OS=="freebsd" or
OS=="openbsd"', {'os_include': 'linux'}],
On 2010/02/04 00:37:26, Evan Martin wrote:
> 80

Done.

http://codereview.chromium.org/565043/diff/1007/1027
File third_party/libxslt/libxslt.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1027#newcode8
third_party/libxslt/libxslt.gyp:8: ['OS=="linux" or OS=="freebsd" or
OS=="openbsd"', {'os_include': 'linux'}],
On 2010/02/04 00:37:26, Evan Martin wrote:
> 80

Done.

[Evan Martin, 2010-02-04 01:54:31]

LGTM

http://codereview.chromium.org/565043/diff/1007/1029
File third_party/ffmpeg/ffmpeg.gyp (right):

http://codereview.chromium.org/565043/diff/1007/1029#newcode50
third_party/ffmpeg/ffmpeg.gyp:50: ['(OS!="linux" and OS!="freebsd" and
OS!="openbsd" and OS!="mac") or use_system_ffmpeg!=0', {
On 2010/02/04 01:46:45, Peter Valchev wrote:
> On 2010/02/04 00:37:26, Evan Martin wrote:
> > 80 cols
> 
> embarrassingly i'm not sure how to wrap this line and keep the syntax correct.
> \ & newline doesn't seem to do it and neither does just newline?

Whoops, no, you are correct in how it's written already

[wtc, 2010-02-04 20:17:35]

LGTM.

I hope we can come up with a solution for the repetitions of
    OS == "linux" or OS == "freebsd" or OS == "openbsd"
and its negation.

Just curious: why don't you want to use tcmalloc on
FreeBSD?

http://codereview.chromium.org/565043/diff/6063/3060
File chrome/chrome.gyp (right):

http://codereview.chromium.org/565043/diff/6063/3060#newcode463
chrome/chrome.gyp:463: '../build/linux/system.gyp:gtk',
Perhaps the build/linux directory should be renamed
build/unix or build/posix at some point...

[pvalchev1, 2010-02-04 22:36:28]

On Thu, Feb 4, 2010 at 12:17 PM,  <wtc@chromium.org> wrote:
> LGTM.
>
> I hope we can come up with a solution for the repetitions of
>    OS == "linux" or OS == "freebsd" or OS == "openbsd"
> and its negation.

Those are defined on the command line when calling gyp_chromium and
hence only one of them is defined at a time. We could define another
common variable and use that instead but with the differences with mac
too, I don't know if really will be any clearer (but sure, long term
there is room for improvement here)

> Just curious: why don't you want to use tcmalloc on
> FreeBSD?

It's called linux_use_tcmalloc, we should rename it and then we can
use it on *BSD... but in this CL I removed it for consistency, because
it's not defined in most other places.


> http://codereview.chromium.org/565043/diff/6063/3060
> File chrome/chrome.gyp (right):
>
> http://codereview.chromium.org/565043/diff/6063/3060#newcode463
> chrome/chrome.gyp:463: '../build/linux/system.gyp:gtk',
> Perhaps the build/linux directory should be renamed
> build/unix or build/posix at some point...

Yes

[Evan Martin, 2010-02-04 22:45:28]

On Thu, Feb 4, 2010 at 2:36 PM, Peter Valchev <pvalchev@google.com> wrote:
> On Thu, Feb 4, 2010 at 12:17 PM,  <wtc@chromium.org> wrote:
>> LGTM.
>>
>> I hope we can come up with a solution for the repetitions of
>>    OS == "linux" or OS == "freebsd" or OS == "openbsd"
>> and its negation.
>
> Those are defined on the command line when calling gyp_chromium and
> hence only one of them is defined at a time. We could define another
> common variable and use that instead but with the differences with mac
> too, I don't know if really will be any clearer (but sure, long term
> there is room for improvement here)

Having seen many of your patches now, I've become more convinced we
should have some sort of "unixlike" define.  I had thought before that
we might be better off with just USE_GTK, USE_NSS, etc. but it becomes
increasingly silly with the variety of things we have here:
$ ls -1 *.target.mk
fontconfig.target.mk
freetype2.target.mk
gconf.target.mk
gdk.target.mk
gtkprint.target.mk
gtk.target.mk
nss.target.mk
selinux.target.mk
x11.target.mk

I guess at some point ChromeOS may wanna remove use of fontconfig,
gconf, gtkprint.  Hrmm, that would be an argument of sorts to leave
'em separate.