Checked in sources which are normally autogenerated during Mesa's build process.

Checked in sources which are normally autogenerated during Mesa's build process.

Generating these sources requires the Python bindings for libxml2 to be installed, and that dependency causes significant problems for Chromium's build process. It is not feasible to add a dependency on a Python module requiring native libraries.

Mesa's XML processors require validation against the DTD, and none of Python's built-in XML processors support validation. The only currently maintained validating XML parsers for Python require native libraries. Therefore it is not possible to rewrite Mesa's XML processors in pure Python.

These sources were generated manually on Linux using a new gyp target. They compile and run successfully on Windows and Mac OS.

This patch is not appropriate to upstream to the Mesa tree.

BUG=238755
TBR=marcheu@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=206317

[Ken Russell (switch to Gerrit), 2013-06-14 06:08:27]

Committed patchset #1 manually as r206317 (presubmit successful).

[Ken Russell (switch to Gerrit), 2013-06-14 06:09:31]

FYI.

This CL was already too large to upload successfully, so I'm not sure the patch
set is correct.

[Stephen White, 2013-06-14 11:25:58]

On 2013/06/14 06:09:31, kbr wrote:
> FYI.
> 
> This CL was already too large to upload successfully, so I'm not sure the
patch
> set is correct.

Hi Ken,

Do we have a README_chromium or similar to document the diffs against upstream
Mesa?

[Myles C. Maxfield (Amazon.com), 2013-06-14 19:54:24]

On 2013/06/14 11:25:58, Stephen White wrote:
> On 2013/06/14 06:09:31, kbr wrote:
> > FYI.
> > 
> > This CL was already too large to upload successfully, so I'm not sure the
> patch
> > set is correct.
> 
> Hi Ken,
> 
> Do we have a README_chromium or similar to document the diffs against upstream
> Mesa?

I was about to suggest doing this, since compile-time generation is nice but not
required. It sounds like dtd verification isn't feasible, so this seems
reasonable for now.

I can make a README_chromium; there are only two tiny changes from upstream.

[Ken Russell (switch to Gerrit), 2013-06-14 21:15:01]

On 2013/06/14 19:54:24, Myles C. Maxfield (Amazon.com) wrote:
> On 2013/06/14 11:25:58, Stephen White wrote:
> > On 2013/06/14 06:09:31, kbr wrote:
> > > FYI.
> > > 
> > > This CL was already too large to upload successfully, so I'm not sure the
> > patch
> > > set is correct.
> > 
> > Hi Ken,
> > 
> > Do we have a README_chromium or similar to document the diffs against
upstream
> > Mesa?
> 
> I was about to suggest doing this, since compile-time generation is nice but
not
> required. It sounds like dtd verification isn't feasible, so this seems
> reasonable for now.
> 
> I can make a README_chromium; there are only two tiny changes from upstream.

Yes. The tar archive I sent to Myles last night contains an updated
README.chromium to go along with the DEPS roll.

[Myles C. Maxfield (Amazon.com), 2013-06-14 21:15:45]

On 2013/06/14 19:54:24, Myles C. Maxfield (Amazon.com) wrote:
> On 2013/06/14 11:25:58, Stephen White wrote:
> > On 2013/06/14 06:09:31, kbr wrote:
> > > FYI.
> > > 
> > > This CL was already too large to upload successfully, so I'm not sure the
> > patch
> > > set is correct.
> > 
> > Hi Ken,
> > 
> > Do we have a README_chromium or similar to document the diffs against
upstream
> > Mesa?
> 
> I was about to suggest doing this, since compile-time generation is nice but
not
> required. It sounds like dtd verification isn't feasible, so this seems
> reasonable for now.
> 
> I can make a README_chromium; there are only two tiny changes from upstream.

Actually, it looks like a git log in the mesa/src directory will describe the
changes we've made from upstream (since f2ba7591b1407a7ee9209f842c50696914dc2ded
is a clean 9.0.3 checkout). Is a README_chromium really necessary?

[Ken Russell (switch to Gerrit), 2013-06-17 19:34:41]

On 2013/06/14 21:15:45, Myles C. Maxfield (Amazon.com) wrote:
> On 2013/06/14 19:54:24, Myles C. Maxfield (http://Amazon.com) wrote:
> > On 2013/06/14 11:25:58, Stephen White wrote:
> > > On 2013/06/14 06:09:31, kbr wrote:
> > > > FYI.
> > > > 
> > > > This CL was already too large to upload successfully, so I'm not sure
the
> > > patch
> > > > set is correct.
> > > 
> > > Hi Ken,
> > > 
> > > Do we have a README_chromium or similar to document the diffs against
> upstream
> > > Mesa?
> > 
> > I was about to suggest doing this, since compile-time generation is nice but
> not
> > required. It sounds like dtd verification isn't feasible, so this seems
> > reasonable for now.
> > 
> > I can make a README_chromium; there are only two tiny changes from upstream.
> 
> Actually, it looks like a git log in the mesa/src directory will describe the
> changes we've made from upstream (since
f2ba7591b1407a7ee9209f842c50696914dc2ded
> is a clean 9.0.3 checkout). Is a README_chromium really necessary?

Yes, a README.chromium is needed both to follow current process as well as make
it easier for people to discover the changes that have been made.

I've sent you a snapshot of the suggested README.chromium in a separate email,
so it should be easy for you to incorporate in your patch upgrading Mesa.