abiword wont launch

[perthie]

I repackaged libgsf.tczl and installed it locally. Abiword now works for me.

[Juanito]

Is there a problem with the libgsf extension? It works for me with other apps.

[perthie]

I tried once again to confirm this.

Booted 2.4rc4 with base norestore

Installed Abiword

The files libgsf-1.so.114.x are missing.

Installed my repackaged libgsf.tczl.

Abiword works.

[curaga]

Can't reproduce, everything works for me since the dep update. Are you using a mirror?

[perthie]

Are you using a mirror?

No, and I'm still getting the same negative result on several machines. I guess someone else will have to weigh in with a test run.

[Juanito]

the libgsf.tczl md5sum checks out OK in the repository

[bigpcman]

the libgsf.tczl md5sum checks out OK in the repository

Here's what I get:

"abiword: error while loading shared libraries: libgsf-1.so.114: cannot open shared object file: No such file or directory"

[curaga]

Are you installing to ram? Maybe that is a factor, I used the conventional tcz way, mounting..

Edit: Well, how nice. GNU cpio as in 2.4rc4 segfaults when loading libgsf to ram.



GNU cpio is no longer used, but it should be checked with busybox cpio. And also if it's because of cpio, or the script.

edit 2. Busybox cpio is more verbose about it: short read, I/O error. The libgsf extension is corrupted, and the corruption is in the dev headers, that's why abiword runs when mounted.

[perthie]

Are you installing to ram?

Yes. I thought that was now the conventional way to load tcz's in 2.4.

All I did to fix libgsf was un-squash it (there were some errors) and re-squash it.

[curaga]

Yes. I thought that was now the conventional way to load tcz's in 2.4.

It's still a choice - of load time, ram use, speed after load, plain preference for either. This choice isn't going away anytime soon.

All I did to fix libgsf was un-squash it (there were some errors) and re-squash it.

That would leave the affected files corrupted. If the tcel is still fine, then it can be used, if not, it would need a rebuild.

[perthie]

It's still a choice

Fair enough. But it's important that both methods are tried during the testing of a new package.

That would leave the affected files corrupted

Actually, I believe the corrupted files got dropped during the un-squash phase. So my re-squashed version was clean (but incomplete).

[enoch]

Thanks everyone its fixed!