Home | Docs | Forums | Lists | Bugs | Planet | Store | GMN | Get Gentoo!

Full Text Bug Listing

When building which-2.19 (unstable) with readline-5.2_p7 (stable) we got the
following error message:

xmalloc.h:29:31: error: readline/rlstdc.h: No such file or directory
make[2]: *** [tilde.o] Error 1
make[2]: Leaving directory
`/var/tmp/portage/sys-apps/which-2.19/work/which-2.19/tilde'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory
`/var/tmp/portage/sys-apps/which-2.19/work/which-2.19'
make: *** [all] Error 2

If using the unstable release of readline-5.2_p12-r1 it compiles without
problem.


Reproducible: Always

Steps to Reproduce:
1. If using stable: ACCEPT_KEYWORDS="~amd64" which-2.19
2. If using unstable AND you did not updated readline: emerge which
3. If using unstable and you updated readline before you should't be reading
this ;-).

Actual Results:  
=sys-apps/which-2.19 fails to merge due compilation errors.

Expected Results:  
Compilation successfull and package merged.

This problem is simple, but if you are installing from a stage-1 you must take
care of updating readline before going to stage-3 (just before emerge -e
system).

When I said if you're compiling from a stage-1 I want to say: If you are
compiling from a stage-1 *using* an *unstable* arch (~amd64 for me).

Created an attachment (id=141652) [details]
Updated which-2.19.ebuild

I know that this is trivial to fix, but I uploaded a which-2.19.ebuild with
DEPEND updated. Just for self-training :).

(From update of attachment 141652 [details])
please do not post entire ebuilds as they are hard to read to figure out what
you've actually changed ... post diffs instead

readline-5.2_p7 provides readline/rlstdc.h ... as does readline-5.1 and
readline-4.3 and ...

whatever the reason for your failure, it isnt a DEPEND issue

since you're building from a stage1, you'll have to figure it out and get back
to us

I have done some research about this.

After decompressing the stage1-amd64-2007.0.tar.bz2 I look for readline header
files:

machine ~ # find /home/username/gentoo -name "readline*"
machine ~ #

As you can see there is no readline by default in stage1 nor bootstrap.sh
installs it. Another check to ensure that:

machine ~ # ROOT=/home/username/gentoo emerge -pv readline

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild  N    ] sys-libs/readline-5.2_p12-r1 to /home/username/gentoo/ 2,018 kB

Total: 1 package (1 new), Size of downloads: 2,018 kB
machine ~ #

Since which-2.19 readline is required before compile this version:

machine ~ # mkdir which
machine ~ # tar zxf /usr/portage/distfiles/which-2.16.tar.gz -C which
machine ~ # tar zxf /usr/portage/distfiles/which-2.19.tar.gz -C which
machine ~ # cg rlstdc which/*
0 which/which-2.19/tilde/xmalloc.h 27 #  include "rlstdc.h"
1 which/which-2.19/tilde/xmalloc.h 29 #  include <readline/rlstdc.h>
machine ~ #

I hope this helps.

last i checked, stage1 installs are not supported ... if it breaks, you get to
fix it

final word is of course up to the release team

I don't think you'll get any argument from us, unless it blocks us building
stages later on down the line, of course :P

Hi,

this comment is only for reference.

If you're currently building a ~keyworded stage3 with catalyst, you must take
care of this bug due it will happend to you.

Confirm, I still have this problem installing gentoo from stage1 on x86
compatible  VIA C7. I have just manualy emerged readline and now which is
emerged correctly. This should be fixed, because I could waste the night of
compilation if I have not noticed that my remote installation has stoped.
People just use emerge -e system and think that there should be no problem
during ~6h of installation/compilation. You will log on and see that for most
of the time there was nothing done because at the some beginning there was
error. It has been reported in January, now we have March.

Peter, as you might not have noticed, this bug is not going to be fixed.  We do
not support stage1 installations.

Hi, 

only for clarification but:

Shouldn't this be fixed to avoid compilation errors while using catalyst? Those
who wants to make a stage3 fall into this error too.

*** Bug 213224 has been marked as a duplicate of this bug. ***

*** Bug 213224 has been marked as a duplicate of this bug. ***

As reported in bug 213224, the which-2.19 build failure is causing a failure in
catalyst when building stages using a stable x86 profile.

This bug is now happenning building stable x86 catalyst profiles. Drobbins
asked about reopen this in the duplicated bug entry.

The rlstdc.h include appears to be extraneous. I have tried removing it and
which appears to build fine. This needs more testing to ensure it builds in all
environments, and has not been tested inside catalyst. See attached
which-2.19-r1.ebuild and which-readline-fix.patch. Also note that a rev bump
should probably not be done as this appears to be purely a build fix (did a rev
bump locally only for testing purposes)

Created an attachment (id=146047) [details]
new ebuild that applies build fix

Created an attachment (id=146049) [details]
remove reference to rlstdc.h - seems to fix things.

in tree as sys-apps/which-2.19-r1, not closing since this is assigned to
release.

erm. I didn't actually intend to revbump it since it's a build fix. it's
actually sys-apps/which-2.19

*** Bug 213224 has been marked as a duplicate of this bug. ***

Well, really this should have stayed RESOLVED and the other should have just
been RESOLVED-FIXED.  I'm marking this as "FIXED" since it's fixed in the tree.
 Again, we don't support stage1 installation, and a build failure for catalyst
isn't the release team's "bug" but one in the package, which has already been
fixed... so again... FIXED.

please properly document patches before putting them in the tree

has anyone sent this to the which maintainer ?

Yes, it has been sent upstream to the which maintainer and he had already
received an identical patch from someone else:

""

This was already sent to me by someone else, and has already been
committed to SVN. It will be in 2.20 -- though it can take years
before I release that :p
- Show quoted text -

On Thu, Mar 13, 2008 at 12:55:04PM -0600, Daniel Robbins wrote:
> Hi Carlo,
>
>  I have created a patch for which-2.19 that removes an unnecessary
>  dependency on readline for compilation.
>
>  While nearly all Linux systems have readline installed, this
>  unnecessary dependency on readline can create problems when
>  "bootstrapping" or building up a Linux OS from scratch, during which
>  time readline might not yet be installed.
>
>  I thought the best course of action was remove the readline dependency
>  - please consider integrating this into your source tree.
>
>  This fix resolves the following Gentoo bugs, which also provide detail
>  of how the issue was breaking Gentoo's distro build process:
>
>  http://bugs.gentoo.org/show_bug.cgi?id=207151
>  http://bugs.gentoo.org/show_bug.cgi?id=213224
>
>  Best Regards,
>
>
>  Daniel Robbins



--
Carlo Wood <carlo@alinoe.com>

""

thanks