Add GN bootstrap step to continuous integration

Add GN bootstrap step to continuous integration

See https://groups.google.com/a/chromium.org/d/msg/chromium-dev/diJmu5aIq-Q/9cvtst5MOx4J for discussion.

BUG=none

[Paweł Hajdan Jr., 2015-06-02 13:44:46]

phajdan.jr@chromium.org changed reviewers:
+ dpranke@chromium.org

[Paweł Hajdan Jr., 2015-06-02 13:44:46]

If you'd like to see this in action the log is
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...

It adds 2 minutes to cycle time, and I'm fine with that from CQ POV.

[Dirk Pranke, 2015-06-02 18:40:34]

dpranke@chromium.org changed reviewers:
+ brettw@chromium.org

[Dirk Pranke, 2015-06-02 18:42:55]

+brettw to get his approval as well.

The change to enable this looks correct.

I have mixed feelings about whether this should be running on the main bots,
given that 99% of chromium devs don't care about this.

Something like the GN upload bots might be more appropriate, but those don't
trigger on every CL.

At some point I'd like to have a dedicated standalone GN repo and bots for it; I
think those would be the right place to run this.

However, I guess in the meantime I think it's okay to run them here.

Brett?

[brettw, 2015-06-03 17:55:02]

I'm concerned about running this on the waterfall. The bootstrap script is a
thing we have in the tree as a service to others that want it, but this isn't
something that's officially supported. I'm reluctant to say that the full Chrome
team should officially support this script since there's no value to those
developers.

Also, I'm concerned that any failures here will be very unclear. It's not clear
how to go from some linker error in a script you've never heard of to needing to
update a Python script when you update base.gypi. Given the low-to-zero benefit
to the team, I don't think the benefits outweigh the costs.

We could have an FYI bot or something, but that seems like a waste of time since
nobody will look at it.

[jochen (gone - plz use gerrit), 2017-02-21 08:14:28]

jochen@chromium.org changed reviewers:
+ jochen@chromium.org

[jochen (gone - plz use gerrit), 2017-02-21 08:14:29]

The script keeps breaking. Can we please reconsider this CL? Sounds like making
this an isolated script should make sure this doesn't slow down the waterfall.

The script could just output a clear message how to fix any failure that occurs,
making it easy for anybody to take action (especially since it will break for
ppl that also modified the base library)

[brettw, 2017-02-21 17:08:07]

How is the bootstrap script breaking affecting you?

Brett

On Tue, Feb 21, 2017 at 12:14 AM, <jochen@chromium.org> wrote:

> The script keeps breaking. Can we please reconsider this CL? Sounds like
> making
> this an isolated script should make sure this doesn't slow down the
> waterfall.
>
> The script could just output a clear message how to fix any failure that
> occurs,
> making it easy for anybody to take action (especially since it will break
> for
> ppl that also modified the base library)
>
> https://codereview.chromium.org/1149413005/
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Dirk Pranke, 2017-02-21 17:11:58]

On 2017/02/21 08:14:29, jochen wrote:
> The script keeps breaking. Can we please reconsider this CL? Sounds like
making
> this an isolated script should make sure this doesn't slow down the waterfall.
> 
> The script could just output a clear message how to fix any failure that
occurs,
> making it easy for anybody to take action (especially since it will break for
> ppl that also modified the base library)

Minor quibble: This is a local script test, not an isolated script test. The
latter
kind runs under swarming, and needs to be completely self-contained. 

We can't easily do that in this case since we need to do full builds, so we'd 
need the toolchains, full copies of all of the needed sources, etc.

[jochen (gone - plz use gerrit), 2017-02-21 19:27:05]

On 2017/02/21 at 17:08:07, brettw wrote:
> How is the bootstrap script breaking affecting you?
> 

For platforms we don't provide binaries for but v8 supports eg ppc or s390
embedders depend on this script

[brettw, 2017-02-21 20:33:54]

This script is never actually used by Chromium developers. Adding this
would be a tax on Chromium developers for the benefit of a very small group
of people like S390 developers (who probably work at IBM on this for their
actual job!). If this was something obvious that fails when you write code,
it might be OK, but this one is particularly obscure: there is a completely
separate build in a different build system that most people don't know
exists.

If we did this, the net result will be that all changes to the base build
file require an extra try run (assuming nobody will remember to update it
the first time). I believe the failure will also be obscure and confusing.
You'll see an undefined symbol "when I already updated the build!" on some
step you don't understand on some bot you've never heard of. Even if we try
to make this clear, it will be quite confusing. And saying people need to
maintain two build systems after we spent so long trying to get off of two
for the benefit of another for-profit company maintaining something an
obscure architecture is too much.

I think if some people care about this, those people should maintain an FYI
waterfall builder themselves.

Brett

On Tue, Feb 21, 2017 at 11:27 AM, <jochen@chromium.org> wrote:

> On 2017/02/21 at 17:08:07, brettw wrote:
> > How is the bootstrap script breaking affecting you?
> >
>
> For platforms we don't provide binaries for but v8 supports eg ppc or s390
> embedders depend on this script
>
> https://codereview.chromium.org/1149413005/
>

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.