Bugzilla Bug 71183

Compilation fails with (from memory) redeclaration of snprintf.
System is a pentium4 with NPTL.



Reproducible: Always
Steps to Reproduce:
1.
2.
3.

Created an attachment (id=43940) [edit]
ebuild for helium with gcc34

The fix is made with a sed line in the ebuild, since the file is created by
configure.
Unfortunately there is problems loading files from the book
Haskell: The craft of functional computing, even though both the book and
helium homepage claims that it works.
Additional patch is commented out.

Created an attachment (id=43941) [edit]
patch to use simple libraries

With this patch helium uses the simple version of libraries, which allows for
loading material from textbook, but alas without overloading, so only current
document is available.

is this patch already in portage? if no, is this the problem you are speaking
about, which the patch should fix or is this something different?

****************************************************
  LVMRUN: the bytecode interpreter
****************************************************
cd ../../lvm/src/runtime; make
make[1]: Entering directory
`/var/tmp/portage/helium-1.2/work/helium-1.2/lvm/src/runtime'
gcc -o core/main.o -c core/main.c -O2 -fno-defer-pop -Wall -I../config -I.
-Icommon
In file included from common/config.h:29,
                 from common/mlvalues.h:28,
                 from core/main.c:15:
../config/s.h:17:1: warning: "strncpy" redefined
In file included from /usr/include/string.h:417,
                 from core/main.c:12:
/usr/include/bits/string2.h:761:1: warning: this is the location of the
previous definition
In file included from common/mlvalues.h:29,
                 from core/main.c:15:
common/misc.h:80: error: conflicting types for 'snprintf'
common/misc.h:80: error: conflicting types for 'snprintf'
common/misc.h:88: error: conflicting types for 'vsnprintf'
common/misc.h:88: error: conflicting types for 'vsnprintf'
make[1]: *** [core/main.o] Error 1
make[1]: Leaving directory
`/var/tmp/portage/helium-1.2/work/helium-1.2/lvm/src/runtime'
make: *** [lvmrun] Error 2

Yes this is the problem that the sed line is for (see ebuild)
configure fails to recognise that (v)snprintf is available.
ofcause fixing configure is the way to go.

The patch kept being rejected when I tried to apply it (manually or with an
ebuild), so I applied the changed manually and made my own patch
(${P}-fixmakefile.patch) and ebuild (the first ebuild posted doesn't apply the
patch, hence the need for one that does)

You should probably test that my patch has the same result as applying the
original patch manually did though

Created an attachment (id=44100) [edit]
Patch to fix the helium-1.2 makefile

Created an attachment (id=44101) [edit]
helium-1.2 ebuild to apply fixmakefile patch

Hi!

Here I submit 'my' :-) patches which (they are just differently named:

1) helium-1.2-install.patch which patches Makefile for installing simple libraries and

2)helium-1.2.ebuild.patch which patches helium-1.2.ebuild

The patches apply cleanly, but I still cannot compile Helium on amd64 - ah, yes, the ebuild patch has keyword for 'amd64' arch.

Sincerely,
Gour

Created an attachment (id=44109) [edit]
patch for patching Makefile.in to be put in ../files folder

Created an attachment (id=44110) [edit]
patch for patching helium-1.2.ebuild

I've done this just now. The gcc-patch is okay, but I didn't apply the
Makefile patch literally. Instead, I modified the ebuild to install both
sets of libraries (first: example-compatible, without type classes, default;
second: limited type class support). The second set of libraries can be
used by calling the helium-tc, texthint-tc, or hint-tc variants of the 
programs.

The new version does not yet include any fixes for amd64.

ks

Some news:

I've reproduced the helium build error on amd64, and I've discussed it
with Arjan and Daan. Daan had discovered a potential problem source quickly,
but that turned out not to be the culprit. He promised to investigate
further ...

ks

Hi!

>I've reproduced the helium build error on amd64, and I've discussed it
with Arjan and Daan. Daan had discovered a potential problem source quickly,
but that turned out not to be the culprit. He promised to investigate
further ...

Thanks a lot for your support.

Sincerely,
Gour

This looks like its been fixed for quite some time now.