Linux: move hardcoded paths to GYP variables.

Linux: move hardcoded paths to GYP variables.

Currently we hardcode several paths in the source (the location of the
sandbox binary and the location of the chrome binary). Packagers are
patching the source to change these, which obviously isn't great.

This patch makes the paths GYP variables, which can be set with -D on
the GYP command line.

[Michael Moss, 2009-07-15 21:19:49]

Would it make sense to make the compile of chrome_sandbox conditional on
linux_sandbox_path being defined? The sandbox binary won't be used without it,
so it offers an extra check in case we forget to set the GYP variable (I'm
assuming that, for instance, packaging will barf on the missing file, as a crude
form of sanity checking). Otherwise, there's no clear indication that you're
running without sandboxing, is there?

http://codereview.chromium.org/149689/diff/1006/8
File chrome/browser/zygote_host_linux.cc (right):

http://codereview.chromium.org/149689/diff/1006/8#newcode48
Line 48: #if !defined(GOOGLE_CHROME_BUILD)
Not sure why this needs to be only !GOOGLE_CHROME_BUILD. It's possible to create
"official" builds for development, in fact, that's my default mode since I
usually want that for testing packaging changes. Is there any reason why this
couldn't be used for such builds?

[agl, 2009-07-15 22:39:05]

On 2009/07/15 21:19:49, mmoss wrote:
> Would it make sense to make the compile of chrome_sandbox conditional on
> linux_sandbox_path being defined? The sandbox binary won't be used without it,
> so it offers an extra check in case we forget to set the GYP variable (I'm

I agree. However, I can't get GYP to do it. If you can, please feel free.

> Not sure why this needs to be only !GOOGLE_CHROME_BUILD. It's possible to
create
> "official" builds for development, in fact, that's my default mode since I
> usually want that for testing packaging changes. Is there any reason why this
> couldn't be used for such builds?

Ok, reverted that part of the change.

Cheers

[Michael Moss, 2009-07-15 22:53:36]

On 2009/07/15 22:39:05, agl wrote:
> On 2009/07/15 21:19:49, mmoss wrote:
> > Would it make sense to make the compile of chrome_sandbox conditional on
> > linux_sandbox_path being defined? The sandbox binary won't be used without
it,
> > so it offers an extra check in case we forget to set the GYP variable (I'm
> 
> I agree. However, I can't get GYP to do it. If you can, please feel free.

I was thinking conditionally include sandbox.gyp in the other gyp file(s) where
it's referenced, but I'll have to look closer to see if that breaks anything
else. No worries, though. Don't let it hold up the CL.