tech-pkg archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Expanding the default OPSYSVARS



Hi Jonathan,

I'd like us to do this properly, and we're very near the start of the
freeze - as you note, there are a number of these changes, so
depending on the successful results of your bulk build, please
consider this a go-ahead for making the OPSYSVARS for PKG_OPTIONS
infrastructure change during the freeze.

Thanks,
Alistair

On 18 March 2016 at 02:46, Jonathan Perkin <jperkin%joyent.com@localhost> wrote:
> * On 2016-03-18 at 04:07 GMT, Darrin B. Jewell wrote:
>
>> Comment below about PKG_SUGGESTED_OPTIONS.${OPSYS} not working correctly below.
>>
>> David Holland <dholland-pkgtech%netbsd.org@localhost> writes:
>>
>> > On Mon, Feb 22, 2016 at 03:24:31PM +0000, Jonathan Perkin wrote:
>> >  > with 'SUBST_CLASSES' listed in OPSYSVARS, one only needs to do this:
>> >  >
>> >  >   SUBST_CLASSES.Darwin+= foo
>> >  >
>> >  > I think one of the reasons it gets so little use is that the default
>> >  > list of variables is too short.  I'd like to add the following
>> >  > variables to the default list, based on some quick analysis of the
>> >  > most-used variables (those used >30 times) within .if ${OPSYS} blocks:
>> >  >
>> >  >   BUILD_TARGET
>> >  >   BUILDLINK_TRANSFORM
>> >  >   CONFIGURE_ARGS
>> >  >   CONFIGURE_ENV
>> >  >   MAKE_ENV
>> >  >   PKG_SUGGESTED_OPTIONS
>> >  >   SUBST_CLASSES
>> >  >   USE_TOOLS
>> >
>> > yes please :-)
>> >
>> > I'd add PKG_SUPPORTED_OPTIONS to that list - doesn't appear that often
>> > but when it does parallelism with PKG_SUGGESTED_OPTIONS will improve
>> > clarity.
>> >
>>
>> Unfortunately, PKG_SUGGESTED_OPTIONS isn't working correctly as an OPSYSVAR
>> A value placed in PKG_SUGGESTED_OPTIONS.Darwin for example does end up
>> in PKG_SUGGESTED_OPTIONS, but it doesn't get correctly propagated to PKG_OPTIONS
>> so the packages don't actually pick up the option.
>>
>> I'm not sure what the right fix is.
>
> Sorry, I thought I'd tested this pretty thoroughly but obviously
> missed the fact that we need the variables evaluated before we can
> check PKG_OPTIONS, so lots is broken by this.  Can you try:
>
>   https://github.com/joyent/pkgsrc/commit/18e6b1a
>
> This seems to work fine for my test package (devel/boehm-gc on
> Darwin), and I'm pushing this through a bulk build test too.
>
> Alistair, heads up that we may need this during the freeze
> unfortunately :(  Alternatively we back out any changes to convert
> PKG_*OPTIONS to OPSYSVARS, though there are quite a few.
>
> --
> Jonathan Perkin  -  Joyent, Inc.  -  www.joyent.com
>


Home | Main Index | Thread Index | Old Index