[gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Donnie Berkholz]

On 12:35 Mon 24 Sep     , Marijn Schouten (hkbst) wrote:
> hkbst       07/09/24 12:35:11
> 
>   Modified:             ChangeLog
>   Added:                bigloo-3.0b_p2.ebuild
>   Log:
>   bump 3.0b-2, minor bugfix version
>   (Portage version: 2.1.3.9)

> # "make test" does something weird so default src_test() in /usr/lib/portage/bin/ebuild.sh fails the following test
> # elif emake -j1 test -n &> /dev/null; then
> # so copy straight from default src_test() all the stuff which depends on that test passing
> src_test() {
> 	vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
> 	if ! emake -j1 test; then
> 		hasq test $FEATURES && die "Make test failed. See above for details."
> 		hasq test $FEATURES || eerror "Make test failed. See above for details."
> 	fi
> }

I'm a bit confused about what's going on here. Isn't src_test() only 
supposed to run when 'test' is in FEATURES?

Thanks,
Donnie
-- 
gentoo-dev@gentoo.org mailing list

[gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Marijn Schouten (hkBst)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Donnie Berkholz wrote:
> On 12:35 Mon 24 Sep     , Marijn Schouten (hkbst) wrote:
>> hkbst       07/09/24 12:35:11
>>
>>   Modified:             ChangeLog
>>   Added:                bigloo-3.0b_p2.ebuild
>>   Log:
>>   bump 3.0b-2, minor bugfix version
>>   (Portage version: 2.1.3.9)
> 
>> # "make test" does something weird so default src_test() in /usr/lib/portage/bin/ebuild.sh fails the following test
>> # elif emake -j1 test -n &> /dev/null; then
>> # so copy straight from default src_test() all the stuff which depends on that test passing
>> src_test() {
>> 	vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
>> 	if ! emake -j1 test; then
>> 		hasq test $FEATURES && die "Make test failed. See above for details."
>> 		hasq test $FEATURES || eerror "Make test failed. See above for details."
>> 	fi
>> }
> 
> I'm a bit confused about what's going on here. Isn't src_test() only 
> supposed to run when 'test' is in FEATURES?

I'm not sure, but as the comment says, I copied this straight from
/usr/lib/portage/bin/ebuild.sh and it goes:

src_test() {
        if emake -j1 check -n &> /dev/null; then
                vecho ">>> Test phase [check]: ${CATEGORY}/${PF}"
                if ! emake -j1 check; then
                        hasq test $FEATURES && die "Make check failed. See
above for details."
                        hasq test $FEATURES || eerror "Make check failed. See
above for details."
                fi
        elif emake -j1 test -n &> /dev/null; then
                vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
                if ! emake -j1 test; then
                        hasq test $FEATURES && die "Make test failed. See
above for details."
                        hasq test $FEATURES || eerror "Make test failed. See
above for details."
                fi
        else
                vecho ">>> Test phase [none]: ${CATEGORY}/${PF}"
        fi
}

Marijn

PS Thank you for reviewing, Donnie.

- --
Marijn Schouten (hkBst), Gentoo Lisp project
<http://www.gentoo.org/proj/en/lisp/>, #gentoo-lisp on FreeNode
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFG+OdHp/VmCx0OL2wRApSBAKCCw8MZPTXuKlbswqYhKxKKrVlggQCgw2FM
OLDCy2QA4DV2tJFED2OFZt4=
=JAnZ
-----END PGP SIGNATURE-----
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Bo Ørsted Andresen]

[-- Attachment #1: Type: text/plain, Size: 1380 bytes --]

On Tuesday 25 September 2007 12:47:35 Marijn Schouten (hkBst) wrote:
> > > # "make test" does something weird so default src_test() in /usr/lib/portage/bin/ebuild.sh fails the following test
> > > # elif emake -j1 test -n &> /dev/null; then
> > > # so copy straight from default src_test() all the stuff which depends on that test passing
> > > src_test() {
> > >     vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
> > >     if ! emake -j1 test; then
> > >         hasq test $FEATURES && die "Make test failed. See above for details."
> > >         hasq test $FEATURES || eerror "Make test failed. See above for details."
> > >     fi 
> > > }
> >
> > I'm a bit confused about what's going on here. Isn't src_test() only
> > supposed to run when 'test' is in FEATURES?
>
> I'm not sure, but as the comment says, I copied this straight from
> /usr/lib/portage/bin/ebuild.sh and it goes:
[SNIP]

Some of that code is unreachable since src_test is never called when test is
not in FEATURES. Also there is some dispute as to whether testing the FEATURES
variable at all is permitted in ebuilds (bug #174335). In either case it's not
needed here. if `emake -j1 test -n` really fails for some reason even though
`emake -j1 test` generally works your src_test should just be:

src_test() {
    emake -j1 test || die "Make test failed"
}

-- 
Bo Andresen

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 1608 bytes --]

On Tuesday 25 September 2007, Bo Ørsted Andresen wrote:
> On Tuesday 25 September 2007 12:47:35 Marijn Schouten (hkBst) wrote:
> > > > # "make test" does something weird so default src_test() in
> > > > /usr/lib/portage/bin/ebuild.sh fails the following test # elif emake
> > > > -j1 test -n &> /dev/null; then
> > > > # so copy straight from default src_test() all the stuff which
> > > > depends on that test passing src_test() {
> > > >     vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
> > > >     if ! emake -j1 test; then
> > > >         hasq test $FEATURES && die "Make test failed. See above for
> > > > details." hasq test $FEATURES || eerror "Make test failed. See above
> > > > for details." fi
> > > > }
> > >
> > > I'm a bit confused about what's going on here. Isn't src_test() only
> > > supposed to run when 'test' is in FEATURES?
> >
> > I'm not sure, but as the comment says, I copied this straight from
> > /usr/lib/portage/bin/ebuild.sh and it goes:
>
> [SNIP]
>
> Some of that code is unreachable since src_test is never called when test
> is not in FEATURES. Also there is some dispute as to whether testing the
> FEATURES variable at all is permitted in ebuilds (bug #174335). In either
> case it's not needed here. if `emake -j1 test -n` really fails for some
> reason even though `emake -j1 test` generally works your src_test should
> just be:
>
> src_test() {
>     emake -j1 test || die "Make test failed"
> }

why is src_test() defined at all ... the default src_test will execute `emake 
check` if possible and then try `emake test`
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Alec Warner]

On 9/24/07, Donnie Berkholz <dberkholz@gentoo.org> wrote:
> On 12:35 Mon 24 Sep     , Marijn Schouten (hkbst) wrote:
> > hkbst       07/09/24 12:35:11
> >
> >   Modified:             ChangeLog
> >   Added:                bigloo-3.0b_p2.ebuild
> >   Log:
> >   bump 3.0b-2, minor bugfix version
> >   (Portage version: 2.1.3.9)
>
> > # "make test" does something weird so default src_test() in /usr/lib/portage/bin/ebuild.sh fails the following test
> > # elif emake -j1 test -n &> /dev/null; then
> > # so copy straight from default src_test() all the stuff which depends on that test passing
> > src_test() {
> >       vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
> >       if ! emake -j1 test; then
> >               hasq test $FEATURES && die "Make test failed. See above for details."
> >               hasq test $FEATURES || eerror "Make test failed. See above for details."
> >       fi
> > }
>
> I'm a bit confused about what's going on here. Isn't src_test() only
> supposed to run when 'test' is in FEATURES?

Yes, but this is done python side (not bash side).

You should not use $FEATURES in an ebuild as they are not meant to be
part of the API.

>
> Thanks,
> Donnie
> --
> gentoo-dev@gentoo.org mailing list
>
>
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Marijn Schouten (hkBst)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Mike Frysinger wrote:
> On Tuesday 25 September 2007, Bo �rsted Andresen wrote:
>> On Tuesday 25 September 2007 12:47:35 Marijn Schouten (hkBst) wrote:
>>>>> # "make test" does something weird so default src_test() in
>>>>> /usr/lib/portage/bin/ebuild.sh fails the following test # elif emake
>>>>> -j1 test -n &> /dev/null; then
>>>>> # so copy straight from default src_test() all the stuff which
>>>>> depends on that test passing src_test() {
>>>>>     vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
>>>>>     if ! emake -j1 test; then
>>>>>         hasq test $FEATURES && die "Make test failed. See above for
>>>>> details." hasq test $FEATURES || eerror "Make test failed. See above
>>>>> for details." fi
>>>>> }
>>>> I'm a bit confused about what's going on here. Isn't src_test() only
>>>> supposed to run when 'test' is in FEATURES?
>>> I'm not sure, but as the comment says, I copied this straight from
>>> /usr/lib/portage/bin/ebuild.sh and it goes:
>> [SNIP]
>>
>> Some of that code is unreachable since src_test is never called when test
>> is not in FEATURES. Also there is some dispute as to whether testing the
>> FEATURES variable at all is permitted in ebuilds (bug #174335). In either
>> case it's not needed here. if `emake -j1 test -n` really fails for some
>> reason even though `emake -j1 test` generally works your src_test should
>> just be:
>>
>> src_test() {
>>     emake -j1 test || die "Make test failed"
>> }
> 
> why is src_test() defined at all ... the default src_test will execute `emake 
> check` if possible and then try `emake test`
> -mike

I cannot reproduce the failure that caused me to copy part of the code from
usr/lib/portage/bin/ebuild.sh so I've removed src_test from this bigloo ebuild
and rely now on the default.

Who's gonna fix usr/lib/portage/bin/ebuild.sh?

Marijn

- --
Marijn Schouten (hkBst), Gentoo Lisp project
<http://www.gentoo.org/proj/en/lisp/>, #gentoo-lisp on FreeNode
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFG+nRpp/VmCx0OL2wRAgszAKCibdBHe5WoBZLprKTSxZ02wrKc0wCeNUwq
9snyDtGguXvZoCkF5p1F5tQ=
=yLpT
-----END PGP SIGNATURE-----
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 141 bytes --]

On Wednesday 26 September 2007, Marijn Schouten (hkBst) wrote:
> Who's gonna fix usr/lib/portage/bin/ebuild.sh?

what's broken in it ?
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 827 bytes --]

[gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Ryan Hill]

Alec Warner wrote:
> You should not use $FEATURES in an ebuild as they are not meant to be
> part of the API.

Can someone put something in the dev guide about this?  I was
looking for exactly this answer a week or so ago and couldn't find
anything. ;)


-- 
                  fonts / wxWindows / gcc-porting / treecleaners
  9B81 6C9F E791 83BB 3AB3  5B2D E625 A073 8379 37E8 (0x837937E8)

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Marijn Schouten (hkBst)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Mike Frysinger wrote:
> On Wednesday 26 September 2007, Marijn Schouten (hkBst) wrote:
>> Who's gonna fix usr/lib/portage/bin/ebuild.sh?
> 
> what's broken in it ?
> -mike

Well, apparently it checking FEATURES is illegal, plus it seems that it
contains dead code. Why not replace it with:


src_test() {
        if emake -j1 check -n &> /dev/null; then
                vecho ">>> Test phase [check]: ${CATEGORY}/${PF}"
                emake -j1 check || die "Make check failed. See above for details."
        elif emake -j1 test -n &> /dev/null; then
                vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
                emake -j1 test || die "Make test failed. See above for details."
        else
                vecho ">>> Test phase [none]: ${CATEGORY}/${PF}"
        fi
}

or in patch version:

- --- /usr/lib/portage/bin/ebuild.sh      2007-09-09 12:50:51.000000000 +0200
+++ /home/marijn/ebuild.sh      2007-09-27 11:58:18.000000000 +0200
@@ -666,16 +666,10 @@
 src_test() {
        if emake -j1 check -n &> /dev/null; then
                vecho ">>> Test phase [check]: ${CATEGORY}/${PF}"
- -               if ! emake -j1 check; then
- -                       hasq test $FEATURES && die "Make check failed. See
above for details."
- -                       hasq test $FEATURES || eerror "Make check failed. See
above for details."
- -               fi
+        emake -j1 check || die "Make check failed. See above for details."
        elif emake -j1 test -n &> /dev/null; then
                vecho ">>> Test phase [test]: ${CATEGORY}/${PF}"
- -               if ! emake -j1 test; then
- -                       hasq test $FEATURES && die "Make test failed. See
above for details."
- -                       hasq test $FEATURES || eerror "Make test failed. See
above for details."
- -               fi
+        emake -j1 test || die "Make test failed. See above for details."
        else
                vecho ">>> Test phase [none]: ${CATEGORY}/${PF}"
        fi


Marijn

- --
Marijn Schouten (hkBst), Gentoo Lisp project
<http://www.gentoo.org/proj/en/lisp/>, #gentoo-lisp on FreeNode
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFG+38Qp/VmCx0OL2wRApdnAJ9Yos8HKuhSX3UDS8rEe3qEeU7AbgCgxlOG
U8eATzDOZf7tFXXnJUD7vZk=
=4TEF
-----END PGP SIGNATURE-----
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 677 bytes --]

On Thursday 27 September 2007, Marijn Schouten (hkBst) wrote:
> Mike Frysinger wrote:
> > On Wednesday 26 September 2007, Marijn Schouten (hkBst) wrote:
> >> Who's gonna fix usr/lib/portage/bin/ebuild.sh?
> >
> > what's broken in it ?
> > -mike
>
> Well, apparently it checking FEATURES is illegal

you misinterpreted the statement ... *ebuilds* using FEATURES is illegal, 
portage itself is 100% free to do it

> plus it seems that it contains dead code. Why not replace it with:

i believe this stems back to src_test always being called irregardless of user 
input ... not sure if that's applicable anymore (so you'd be right, there'd 
be no point)
-mike

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Marijn Schouten (hkBst)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Mike Frysinger wrote:
> On Thursday 27 September 2007, Marijn Schouten (hkBst) wrote:
>> Mike Frysinger wrote:
>>> On Wednesday 26 September 2007, Marijn Schouten (hkBst) wrote:
>>>> Who's gonna fix usr/lib/portage/bin/ebuild.sh?
>>> what's broken in it ?
>>> -mike
>> Well, apparently it checking FEATURES is illegal
> 
> you misinterpreted the statement ... *ebuilds* using FEATURES is illegal, 
> portage itself is 100% free to do it

I think default ebuild functions should also be legal when implemented in an
actual ebuild and thus I don't think default ebuild functions should be
considered to be part of portage.

There is no reason to allow defaults to do special magic stuff and it is
unclean to allow it. Any special stuff should either be made part of the API
so all ebuild functions can use it or removed from defaults and normal
functions alike.

Marijn

- --
Marijn Schouten (hkBst), Gentoo Lisp project
<http://www.gentoo.org/proj/en/lisp/>, #gentoo-lisp on FreeNode
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFG+7FPp/VmCx0OL2wRAj5wAJ4yYSKCwlY/Skj6/Xg+eKKpnEe2GQCfXBSb
70i45u3RdXr11fFDc1C+nDw=
=oMWg
-----END PGP SIGNATURE-----
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-commits] gentoo-x86 commit in dev-scheme/bigloo: ChangeLog bigloo-3.0b_p2.ebuild

[Jeroen Roovers]

On Wed, 26 Sep 2007 19:01:33 -0600
Ryan Hill <dirtyepic@gentoo.org> wrote:

> Can someone put something in the dev guide about this?  I was
> looking for exactly this answer a week or so ago and couldn't find
> anything. ;)

Maybe it's a clue that [1] does /not/ mention FEATURES as a variable to
be used in ebuilds, but only as a variable that emerge might use.


Kind regards,
     JeR



[1] http://devmanual.gentoo.org/ebuild-writing/variables/index.html
-- 
gentoo-dev@gentoo.org mailing list