Last day to vote/reject on proposed EOL names

[Talden]

> I was assuming the rules for EOL handling would be stored, eventually, in
> the repo, so everyone has it.  Further, it is part of the core, so there
> should be no chance for the 'hooks' to not be installed.  IOW, I'm arguing
> the hooks for rejection would be an integral part of EOL support.

Ahh that's not how I interpreted it... Sorry, my English isn't very
good (I wish I knew another language...).

> Could you sell that bzr will refuse to allow incorrect line endings, or is
> the only thing you can sell that bzr will "silently ignore them and do the
> right thing"?

Actually I think that would be fine.  It would need to do this much
better than Subversion (which busts on the first violation only making
it a "many attempts" problem if you have several).  It would be
trivial for us to include EOL correcting scripts to run when this
eventuates.  Silently ignoring could, conceivably, do the wrong thing
- throwing up an error sounds like a better choice.

> I think I'm still missing a key part of the picture though, as I'm not sure
> what "the right thing" is in these scenarios (eg, if the user is using a
> brain-dead editor, and it happens to be open while they do an update, and
> the update doesn't give back the mangled thing they initially checked in,
> things are already getting confused...)

I would expect that, if something is already munged and committed that
it must be corrected before the rule can apply.  That is, validating
rule changes would be desireable.

> I'll try and re-read some of the docs in more detail...

Indeed, I think I need to look closer at how this has been implemented
in bazaar.

--
Talden