do symbolic links inside a branch cause problems?
Wichmann, Mats D
mats.d.wichmann at intel.com
Mon Jun 25 16:54:46 UTC 2012
I have a couple of cases where branches that have symbolic
links inside them seem to cause problems for certain ops.
I've never really gotten a good feel for why this is an issue,
so I'm wondering if there's any known limitations in this area?
I have a vague memory of reading something but can't find
anything relevant with some searches.
the branches in question are used as local mirrors of
upstream to mitigate a slow internet connection, and don't
ever get any local changes, but they do need to be kept
in sync with the masters. bzr multi-pull has problems on two
of these branches, reporting a failure and quitting, while
a plain pull in those branches don't show a problem, which
kind of indicates there are different code paths being taken
in updating the working tree.
I just tried the bzr list-branches command in these and
it tips over the same way (Transport error: [Errno 40]
Too many levels of symbolic links)
Indicating the link is being followed, when it probably shouldn't.
In both instances, the links are to a relative path
FOO -> . (don't ask, this is a portion of a website and apparently
the link is required to make some historical references work)
modules -> ../../.. (this one is devops information and the link has
to do with deployment onto targets, I'm told)
there are other symbolic links in branches that don't cause problems,
it's only these where the target is a "." or "..".
Linux host, bzr 2.5.0 (and all earlier versions I've tried it on)
I'll file a bug if that's what it is, of course.
More information about the bazaar
mailing list