Closed alexshpilkin closed 11 years ago
This is a bit problematic. I have to find a package that does the right thing in this scenario.
As it turns out, there's a dedicated AM_ICONV
macro in iconv.m4 from GNU gettext; you don't have to use gettext itself to utilize it. And it should be possible to draw some inspiration from it anyway: the git people wrote the check themselves, for example.
Thanks! I have included scripts and changed iconv detection in configure.ac
in the head of the devel
branch. It works on my system.
Looks to be fixed now, although I no more see a way to disable iconv support altogether (is it intended?). Unfortunately, #46 has recently been introduced again:
../../doc/stdlib.texi:1830: table requires an argument: the formatter for @item
../../doc/stdlib.texi:3633: table requires an argument: the formatter for @item
../../doc/stdlib.texi:9091: table requires an argument: the formatter for @item
So, iconv should be fine now, but I’m unable to fully verify it.
It should be fine in the head of the devel
branch. Thanks again.
The configure script assumes that the
(vicare iconv)
library can only be built if alibiconv
is available. This is not true as glibc can (and is, in e.g. Arch Linux) be built withiconv()
and related APIs integrated instead, meaning that there is no separatelibiconv
on the system and leading to a spurious failure. I am not familliar enough with autotools to rewrite this myself at the moment, but I figure thatVICARE_WITH_LIBRARY
in its current form cannot support this case.