cl-librarian-devel

Download

cl-librarian-devel@lists.common-lisp.net

October 2008

1 participants
1 discussions

Re: [Cl-librarian-devel] cl-librarian patch
by Maciej Pasternacki 30 Oct '08

30 Oct '08

Hello, Thanks for the patch. I had to apply manually, because either mine, or your mail program mangled some white space, and checksums were wrong. Darcs changesets need to be mailed as attachments, not pasted inline. Unfortunately this means you'll probably have to unrecord your version of patch to pull from main repo, sorry for this. On 2008-10-30, at 14:43, Kazuo Koga wrote: > I wrote some patch to cl-librarian for me. > 1) support Mercurial (hg) Applied and pushed to public repo, thanks. > 2) CVS workaround for my platform (OS X) I didn't apply this yes, mainly because I try, and, up to now, managed to avoid caling shell. That was why I recently backed out of using trivial-shell for running commands. The "cd $DIR && ..." trick is neat, but there are some hidden issues regarding portability. Most important is Windows platform -- some Bourne shell may be downloaded for it, but it's a dependency that's actually unneeded; second one is quoting -- using "cd '~A' ..." is not enough, because of possibility of directory name containing an apostrophe (especially on desktop systems, like Windows and Mac OS X). I thought that passing absolute path to "cvs checkout -d" would be enough, but it is officially unsupported ("for security reasons", as if I didn't know which path I give to checkout), and not documented at all. My fault for not having this tested enough. I will try to find some workaround without using shell, but this may be hard, because using "../"-starting pathnames is also unsupported. If I don't find the real fix, I'll patch it with calling shell, but this will need more work (customizable *shell-command*, quoting parameters, calling cmd.exe on Windows). This might be needed anyway for Git, see http://comments.gmane.org/gmane.comp.version-control.git/98584 (though I'll need to bump this thread to check whether there was any conclusion at all). Regards and thanks, Maciej. -- -><- Maciej Pasternacki -><- http://www.pasternacki.net/ -><-

1 0