Bugzilla Bug 220040

Compilation of bash, gawk (and probably many others) stops for 60 seconds with
message checking for working mktime... and resuls is "no".



Reproducible: Always

Steps to Reproduce:
# emerge -e system
...
# grep grep 'mktime.*no' /var/log/portage/*.log
/Gentoo/var/log/portage/app-shells:bash-3.2_p33:20080502-122356.log:checking
for working mktime... no
/Gentoo/var/log/portage/sys-apps:gawk-3.1.5-r5:20080502-120142.log:checking for
working mktime... no
Actual Results:  
Bundled mktime is used

Expected Results:  
System mktime is used

This is caused by problem in old version of AC_FUNC_MKTIME - autoreconf fixes
this problem nicely but it's not enabled for bash and/or gawk

I recently recompiled world with gcc-4.3 (~1100 packages), and from these the
following ones are concerned, i.e. their log contains "working mktime... no"

app-arch/cabextract-1.2
app-editors/emacs-22.1-r4
app-shells/bash-3.2_p33
app-text/a2ps-4.13c-r6
dev-util/cvs-1.12.12-r4
sys-apps/gawk-3.1.5-r5
sys-libs/gwenhywfar-2.6.1

really should be handled automatically ...

Created an attachment (id=151883) [details]
autoconf-mktime.patch

Created an attachment (id=151885) [details]
autoconf-mktime-2.61.patch

I think solar said he has something of the like, we should probably ask him..
and we might find time next week to coordinate on that? It didn't make SoC so
we'll have to handle it ourselves at this point.

*** Bug 220611 has been marked as a duplicate of this bug. ***

net-ftp/pure-ftpd-1.0.21-r1

Is this bug the right place to report these packages? Nothing has been
commented here for 6 months.

net-fs/curlftpfs-0.9.2
Looks like the same issue.

Hmm. Or not. Sorry, have no idea why but works in 2 min again.

Both dev-libs/pkcs11-helper-1.05 and dev-libs/pkcs11-helper-1.06 are using the
broken version.

media-libs/gmyth is affected by this. If the build thinks mktime is broken,
wouldn't that cause issues is some cases?

Actually gmyth-0.7.1 (not in Portage) doesn't seem to be affected.

Looks like autoconf-2.62 fixed this issue. Could we update eclass to depend to
>=autoconf-2.62 so our eautoreconf fixes started to work? Thanks.