sparse: override dirstate.walk() instead of dirstate._ignore
Instead of treating files that are outside the sparse config as
ignored, this makes it so we list only those that are within the
sparse config by passing the sparse matcher to dirstate.walk().
Once we add support for narrow (sparseness applied to history, not
just working copy), we will need to do a similar restriction of the
walk over manifests, so this will be more consistent then. It also
simplifies the code a bit.
Note that a side-effect of this change is that files outside the
sparse config used to be listed as ignored, but they will now not be
listed at all. This can be seen in the test case where "hg purge" no
longer has any effect because it doesn't see that the files outside
the space config exist. To fix that, I think we should add an option
to dirstate.walk() to walk outside the sparse config. We might expose
that to the user as --no-sparse flag to e.g. "hg status" and "hg
purge", but that's work for another day.
Differential Revision: https://phab.mercurial-scm.org/D59
cHg
===
A fast client for Mercurial command server running on Unix.
Install:
$ make
$ make install
Usage:
$ chg help # show help of Mercurial
$ alias hg=chg # replace hg command
$ chg --kill-chg-daemon # terminate background server
Environment variables:
Although cHg tries to update environment variables, some of them cannot be
changed after spawning the server. The following variables are specially
handled:
* configuration files are reloaded automatically by default.
* CHGHG or HG specifies the path to the hg executable spawned as the
background command server.
The following variables are available for testing:
* CHGDEBUG enables debug messages.
* CHGSOCKNAME specifies the socket path of the background cmdserver.
* CHGTIMEOUT specifies how many seconds chg will wait before giving up
connecting to a cmdserver. If it is 0, chg will wait forever. Default: 60