On Wed, Sep 03, 2008 at 05:33:28PM -0600, Mark J. Nelson wrote: > Will Fiveash wrote: > >> Why? I ask this because this breaks one of hg status useful features >> which is reporting unknown files. And this file is part of the >> repository so if I want to change it or remove it in my repository hg >> status tells me it's modified/removed and I assume that this will cause >> problems with commit/push. > > Because nothing else is both accurate and tolerable on a built > workspace.
Who says one has to build in the repository (think zfs snapshot/clone and rsync)? I think having a .hgignore that ignores all files not explicitly added to the repository is wrong. I like the fact that if I create a file in my repository but forget to do the hg add I'll see: ? usr/src/mynewfile.c when I run hg status. For people that do build in their repository they can always do 'hg st -mard'. And I'll bet money that someone will get confused and use hg rm on .hgignore and push that back. > The extensive pattern space necessary to properly ignore all > of the ON build products drives Mercurial performance to its knees, and > the list of ON build products, proto area contents, logs, archives, and > packages is long enough to make "hg status -u" output unusable. Anything > in between is, to varying degrees, incorrect or intolerable. Fine, don't build in the repository. Or if someone does then they can create their own, global $HOME/.hgignore. > That said, you may "rm .hgignore," but please don't "hg rm .hgignore." > > The former will report .hgignore as > > ! .hgignore > > ...but a subsequent hg commit will NOT attempt to remove the file. kkkkkkkkkkkkk -- Will Fiveash Sun Microsystems Office x64079/512-401-1079 Austin, TX, 78727 (TZ=CST6CDT), USA Internal Solaris Kerberos/GSS/SASL website: http://kerberos.sfbay.sun.com http://opensolaris.org/os/project/kerberos/
