seq is not working.

[rdaneel]

This is the Haiku's seq output: ~> seq 10 0 0 0 0 0 0 4 c2 00

It should be: 1 2 3 4 5 6 7 8 9 10

I dont know where the problema is. The source looks good: ​http://svn.berlios.de/viewcvs/haiku/haiku/trunk/src/bin/coreutils/src/seq.c?rev=19747&view=markup

And this sample program is working. #include <stdio.h> int main() {

int max = 5; long i; for (i=0; i < max ; i++) printf("%L ",i);

}

[axeld]

Ingo and I had investigated the issue a while back, and it looks our printf() is broken for "long long double". Since Korli had last worked on glibc's printf(), I'm assigning it to him - I hope he'll like it :-)

[korli]

Just had a look and the weird thing is : it's working ok with a gcc4 build, but not a gcc2.95.3 ...

[bonefish]

Probably the compiler would need long long double support and gcc 2.95.3 might not implement it (properly), while gcc 4 does. I suppose it's some configuration issue of glibc (or coreutils?), but Axel and I didn't spot it.

[korli]

seq was changed in 6.7 to use long double : ​http://svn.berlios.de/viewcvs/haiku/haiku/trunk/src/bin/coreutils/src/seq.c?rev=19747&r1=17309&r2=19747

using the older version works with gcc2.95.

Could we keep the old copy around to build with gcc 2.95 ?

[korli]

I copied the old seq.c to be used with gcc 2.95 in hrev21883.

[bonefish]

I wonder, if other programs of the coreutils have similar problems. Maybe it would be better to have a complete copy of the old coreutils. Not sure...

[axeld]

I would guess they might also have fixed several vulnerabilities, and also improved the usage of several commands, so I don't think it's a good idea to just remove those changes.

I think replacing seq with an older version is the way to go, though, and we could take a similar route for other apps, but IMO we should only do so when we actually encounter any problems with them.

[korli]

Fixed for real in hrev23127. Reverted hrev21883 in hrev23128.

[korli]

(clearing up) As the fix is for gcc, it should be rebuilt to have the bug fixed.