On 10/01/2011 11:27 AM, Fabian Groffen wrote:
> Hi Zac,
>
> On 01-10-2011 10:34:02 -0700, Zac Medico wrote:
>> As I integrate prefix support into mainline portage, I think it will
>
> Cool! and Thanks!
>
>> make more sense to use $EROOT instead of $ROOT for keys to portage.db
>> and similar map objects. This will also affect the portageq commands
>> which take a <root> parameter. The reason that I think $EROOT makes more
>> sense for these keys is that it will allow for multiple prefixes to
>> exist simultaneously in maps like portage.db.
>>
>> This won't affect non-prefix users, since $EROOT == $ROOT when $EPREFIX
>> is empty. So, I'm asking here because if might affect prefix users who
>> use portageq, or any programs installed in a prefix that use the
>> sys-apps/portage python API. If necessary, I suppose that python
>> programs could have some compatibility code which checks whether or no
>> $EROOT is contained in portage.db, and fall back to "/" otherwise.
>
> What does it actually mean? Does one have to use
> portageq envvar CHOST $EPREFIX/
> instead when this is implemented?
> That would seem not correct to me.
Well, it wouldn't apply to portageq's envvar command, since that doesn't
have a <root> argument. These are the portageq commands that would be
affected:
all_best_visible <root>
best_version <root> <category/package>
best_visible <root> [pkgtype] <atom>
contents <root> <category/package>
expand_virtual <root> <atom>
filter_protected <root>
get_repo_path <root> <repo_id>+
get_repos <root>
has_version <root> <category/package>
is_protected <root> <filename>
list_preserved_libs <root>
mass_best_version <root> [<category/package>]+
mass_best_visible <root> [<category/package>]+
match <root> <atom>
metadata <root> <pkgtype> <category/package> [<key>]+
owners <root> [<filename>]+
--
Thanks,
Zac
RFC: changing sys-apps/portage python API to use $EROOT instead of $ROOT for keys to portage.db and similar map objects
