1.1.3-r2813: include fix for example dir / test-charpoly compile failure without givaro

[mabshoff]

** Issue 1:

test-charpoly.C doesn't compile without Givaro configured in:

g++ -g -DLinBoxSrcOnly -Wall -O2 -o test-commentator test-
commentator.o -L/usr/lib64 -llapack -lblas -lgmp -lgmpxx
if g++ -DHAVE_CONFIG_H -I. -I. -I.. -I.. -I../linbox -g -
DLinBoxSrcOnly -Wall -O2 -MT test-charpoly.o -MD -MP -MF ".deps/test-
charpoly.Tpo" -c -o test-charpoly.o test-charpoly.C; \
then mv -f ".deps/test-charpoly.Tpo" ".deps/test-charpoly.Po";
else rm -f ".deps/test-charpoly.Tpo"; exit 1; fi
In file included from test-charpoly.C:29:
../linbox/ring/givaro-polynomial.h:14:29: error: givaro/givpoly1.h: No
such file or directory
../linbox/ring/givaro-polynomial.h:15:35: error: givaro/
givpoly1factor.h: No such file or directory
In file included from ../linbox/ring/givaro-polynomial.h:18,
from test-charpoly.C:29:
../linbox/element/givaro-polynomial.h:14:29: error: givaro/givpoly1.h:
No such file or directory

I didn't fix this, but there are several possibilities: #ifdef the
test that depend on Giovaro or the in my opinion better solution:
split the test depending on Givaro into its own test and run it
conditionally depending whether Givaro is configured in or not.

** Issue 2:

The issue was reported to me by Stefan Kaspar.

I fixed the issue: The examples needs "-I$(top_srcdir)/linbox" to
compile on RHEL 5/Debian testing:

kaspar@faeles:/usr/local/src/mathlibs/linbox-svn-070907/examples$ make
echelon
g++ -DHAVE_CONFIG_H -I. -I.. -g -DLinBoxSrcOnly -MT echelon.o
-MD
-MP -MF .deps/echelon.Tpo -c -o echelon.o echelon.C
In file included from echelon.C:9:
../linbox/field/modular-double.h:20:27: error: linbox-config.h: Datei
oder
Verzeichnis nicht gefunden
In file included from ../linbox/integer.h:22,
from ../linbox/field/modular-double.h:21,
from echelon.C:9:
../gmp++/gmp++.h:7:27: error: linbox-config.h: Datei oder Verzeichnis
nicht gefunden
make: *** [echelon.o] Fehler 1

The patch is at
http://fsmath.mathematik.uni-dortmund.de/~mabshoff/patches/linbox-1.1.3svn-r2813-fix-include-issue-in-example-directory.patch

Next up: lround() m4 macro.

Cheers,

Michael

[Clement Pernet]

On Sep 8, 9:50 am, mabshoff <Michael.Absh...@fsmath.mathematik.uni-

dortmund.de> wrote:
> ** Issue 1:
>
> test-charpoly.C doesn't compile without Givaro configured in:

Fixed by removing the dependencies on Givaro in test-charpoly.

> ** Issue 2:
>
> The issue was reported to me by Stefan Kaspar.
>
> I fixed the issue: The examples needs "-I$(top_srcdir)/linbox" to
> compile on RHEL 5/Debian testing:

I prefer to change all the includes into
#include "linbox/linbox-config.h"
to keep it consistent with the other includes.

I have applied my fix.
I guess, I will have to revert some recent patch where we added -I$
(top_srcdir)/linbox in many places.
I'll look at it later on.

Michael, please object if you see any other good reason to prefer your
solution.

Clement

[mabshoff]

On Sep 10, 3:37 pm, Clement Pernet <clement.per...@gmail.com> wrote:
> On Sep 8, 9:50 am, mabshoff <Michael.Absh...@fsmath.mathematik.uni-
>

Hello Clement,

> dortmund.de> wrote:
> > ** Issue 1:
>
> > test-charpoly.C doesn't compile without Givaro configured in:
>
> Fixed by removing the dependencies on Givaro in test-charpoly.
>

Okay, great. I will check it out.

> > ** Issue 2:
>
> > The issue was reported to me by Stefan Kaspar.
>
> > I fixed the issue: The examples needs "-I$(top_srcdir)/linbox" to
> > compile on RHEL 5/Debian testing:
>
> I prefer to change all the includes into
> #include "linbox/linbox-config.h"
> to keep it consistent with the other includes.
>
> I have applied my fix.
> I guess, I will have to revert some recent patch where we added -I$
> (top_srcdir)/linbox in many places.
> I'll look at it later on.
>
> Michael, please object if you see any other good reason to prefer your
> solution.
>

Nope, that is fine with me. I just wanted to make it compile with
minimum effort. Now that you mention it I do remember that the plan
was to use '#include "linbox/linbox-config.h"' so I don't mind to much
that you went through the code and fixed all those file manually.

> Clement

Cheers,

Michael

[mabshoff]

<SNIP>

Hello Clement,

> Michael, please object if you see any other good reason to prefer your
> solution.
>

make && make check works fine.

But I was wondering if you could you do separate check ins next time
for two unrelated issues. That would be useful for Sage, ApCoCoALib
and other external users of the libraries if they want to pick out
certain changesets and apply them to an older revision. It just makes
it less likely to have collisions or mismerges.

> Clement

Cheers,

Michael

[Clement Pernet]

On Sep 10, 8:05 pm, mabshoff <Michael.Absh...@fsmath.mathematik.uni-
dortmund.de> wrote:

> Nope, that is fine with me. I just wanted to make it compile with
> minimum effort. Now that you mention it I do remember that the plan
> was to use '#include "linbox/linbox-config.h"' so I don't mind to much
> that you went through the code and fixed all those file manually.

Fortunately, it was not done manually ;)

Sorry for the unique check in. I usually try to do them separately but
forgot it this time.

Clement