[haiku-development] Re: Several script-based errors in the build system
- From: Matt Madia <mattmadia@xxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Mon, 6 Jul 2009 00:50:08 +0000
On Mon, Jul 6, 2009 at 00:15, Ingo Weinhold<ingo_weinhold@xxxxxx> wrote:
>> Thing is, even if one of these errors occur, jam seems to return 0 (no
>> error)
>> Should jam be treating these as an actual error?
>
> Jam is not really to blame. The problem is that in such a case the scripts
> generated by the build system contain scrambled lines. Since the scripts
> don't abort when an error is encountered and the last command is one that
> doesn't fail, jam can't know that something went wrong.
>
Just to make sure, these script errors can result in imagefiles that
contain unexpected or undesirable errors/behaviors?
side note: eventually, i want sikosis to upgrade his machine to use
this script -- once uploading is implemented. so, i'll probably
re-work the detection&handling of these script errors to ensure that
the generated image files do not contain them.
> For the time being the only really reliable method it to build with a single
> job. I suppose for build-o-matic it isn't even a big problem, as you could
> just run the builds for the different target architectures/ABIs in parallel.
>
Merely out of curiosity, do you envision the script to spawn off
threads to compile each target or something else?
--mmadia
Other related posts: