[Sorry if this is a duplicate; my mail server got (incorrectly) placed in a spam blocking list and some fraction of my mail never reached its intended recipient] [Because I seem to be the one that can see these messages; I'm forwarding them to the list]
Begin forwarded message: > From: Marcin Bachry <[email protected]> > Date: April 16, 2009 6:10:22 AM EDT > To: David Abrahams <[email protected]> > Subject: Re: tramp (2.1.16-pre); Needless remote access? > > 2009/4/16 David Abrahams <[email protected]>: >> I'm not sure how this could have happened, but I just found TRAMP >> initiating remote acesses when I was staging/un-staging files in >> magit >> on a *local* git repository. I did (setq default-directory "~") in >> the >> magit buffer to try to stop it, to no avail. > > I can reproduce it by typing "g" (magit-refresh) in magit status > buffer. I think the function "magit-revert-buffers" (run from > "magit-refresh-wrapper") is responsible for the bug: it scans all > buffers (including tramp ones) and does file operation on them > ("verify-visited-file-modtime" function). This wakes up tramp > connections and is generally slow. > > One obvious workaround is to ignore remote buffers using > "buffer-remote-p". But I bet there's a smarter solution... > > m. > Begin forwarded message: > From: Michael Albinus <[email protected]> > Date: April 16, 2009 9:24:04 AM EDT > To: David Abrahams <[email protected]> > Cc: [email protected] > Subject: Re: tramp (2.1.16-pre); Needless remote access? > > David Abrahams <[email protected]> writes: > >>>> I also have been unable to force the problem to be reproduced so >>>> far. >>> >>> No idea about. For further analysis from Tramp pov, I would need a >>> scenario to reproduce it ... >>> >> I think this might explain it: > > Thanks! So I happily ignore it, because it must be fixed in magit. > > Maybe they can implement something like `recentf-keep', which calls > `recentf-keep-default-predicate' with special handling of remote > files. > > Best regards, Michael. > -- David Abrahams BoostPro Computing http://boostpro.com
