Trent W. Buck wrote: > Ben Franksen <[email protected]> writes: >> Ashley Moran wrote: >>> On 4 Apr 2009, at 16:23, Guillaume Hoffmann wrote: >>>> When you edit the long comment with darcs amend-record >>>> --edit-long-comment , the first line is in fact the name (or short >>>> description) of the patch. >>> >>> Thanks! I tested this and it works. >>> >>> I find this really non-intuitive though. I'd really like a way to do >>> inline patch renaming via --edit-description, like the way you first >>> enter a patch name. >> >> I agree. I found this quite unintuitive, even a hassle, when I learned >> darcs. Sometimes you don't have a $DISPLAY in Unix. Then some >> obscure 'default' editor comes up (this depends on system configuration >> and is not always appropriate to change, for instance when using shared >> accounts) which I don't even know how to exit from. > > Arguably learning how to exit without saving from vi and nano should be > one of your first UNIX lessons :-)
I know vi enough to do very simple tasks and doesn't nano list the most important keys at the bottom? However, I once got an emacs as $EDITOR. Tried every combination of keys I could think of but to no avail. IIRC, in the end I did a Ctrl-Z and killed it ;-) > To customize the editor temporarily on a shared account (such as root), > you can set VISUAL and/or EDITOR to the name of your preferred editor, > e.g. > > export EDITOR=ed > darcs record > > Or one-shot > > EDITOR=ed darcs record > > Because it's not appropriate to make this permanent in .profile on a > shared account, what we typically do at my workplace is each have our > own $ENV script in /root, e.g. > > /root/.shrc.twb > /root/.bashrc.alla > /root/.bashrc.russm > > Then when you first log in, you just do > > . ~/.shrc.twb > > to get your preferred environment. Thanks, I know all these tricks. I would have done so had I expected trouble. But I did not. Besides, I am also speaking on behalf of much less knowledgeable *nix users, most of whom really don't want to be bothered with these things. >> Would a patch be accepted that adds another switch to amend-record >> (e.g. --edit-name)? The idea is that this starts only the line editor, >> and presents the old patch name to be edited. > > There is already -m for that purpose, though it requires > copy-and-pasting, I think the current choices of -m and --edit are > adequate for amend-record, but I'm prepared to listen to arguments for a > third intermediary option. Funny, I never realized that -m works with amend-record. Thanks for the tip. > I definitely DO think it's confusing the way all editor sessions > currently have the patch name and the patch description, with no blank > line separating them or other visual clues. I think this is EXTREMELY > confusing for anyone who isn't already coming from a VCS that does that > (e.g. CVS?) > > Even for me, it causes problems because I don't think I should add a > blank line between patch name and description, but if I don't, it causes > wrapping problems for my editor (which thinks it's all a single > paragraph). > > Perhaps a better style would be that of an email message, e.g. > > Author: <patch author> > Date: <date> > Summary: <patch name> > > <patch description> > > This would also provide an alternative way to spoof author and date. > Currently the latter can only be spoofed with --pipe, which is a massive > pain in the arse for interactive use, as input validation isn't done > until the end. Good idea. Cheers Ben _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
