Bug #43181 for CPAN-Site: Indexing fails with long filenames

Tue Feb 10 08:19:19 2009 DTADY [...] cpan.org - Ticket created

CC:	dexterbt1 [...] yahoo.com
Subject:	Indexing fails with long filenames

Hi, We are happily using cpan site as a deployment tool. Recently though, we encountered a bug that fails to properly index our own private modules. The problem lies when the CPAN::Site::Index::inspect_entry() parses the gzip output of the raw tar stream. We are using GNU tar 1.51.1 and very long files are marked as "././\@LongLink"; IIRC, GNU tar uses this to store long filenames that are otherwise limited in older implementations of tar. Please do note that we have a dense namespace tree, and some names become inevitably long, for clarity sake. I apologize as I am unable to attach the distribution that reproduces the bug (NDA). I will be happy though to assist you if you are unable to reproduce the bug yourself. Thank you. Dexter Tad-y

Thu Feb 26 05:43:03 2009 MARKOV [...] cpan.org - Correspondence added

On Tue Feb 10 08:19:19 2009, DTADY wrote: Show quoted text

> The problem lies when the CPAN::Site::Index::inspect_entry() parses the > gzip output of the raw tar stream. We are using GNU tar 1.51.1 and very > long files are marked as "././\@LongLink"; IIRC, GNU tar uses this to > store long filenames that are otherwise limited in older i

I do not know whether this is simple to fix in the current code. Different tar implementations do different things when filenames exceed 100 characters. You may experiment with it yourself... On the moment, I am working to change the way CPAN::Site works, more into how CPAN itself works: cleanly unpack the tar archieves. That will solve your problem for sure.

Thu Feb 26 05:43:03 2009 The RT System itself - Status changed from 'new' to 'open'

Sun Sep 06 18:15:58 2009 MARKOV [...] cpan.org - Correspondence added

CPAN::Site rewriten to use Archive::Tar. This probably solves the problem.

Sun Sep 06 18:16:02 2009 MARKOV [...] cpan.org - Status changed from 'open' to 'resolved'