On 12/15/05, Juliusz Chroboczek <[EMAIL PROTECTED]> wrote:
> Sorry, it needs more work.  I know it's annoying, but that's important
> missing functionality you're implementing, and I'd like it to be just
> right.

No problem.  I'll do as many revisions as it takes until everyone is
happy.  I'm still learning Haskell and the Darcs code, so I expect the
need for corrections.

>
> > +                             where f = ".darcs-record"
>
> There's no reason why that should be a hidden file (one that starts
> with a dot); quite the opposite, it should be a file that newbies will
> notice straight away.
>
> The trouble is, if you use a normal file, you'll run the risk of
> clobbering a user's file.  So you'll need to take some precautions
> against that.  The function Lock.world_readable_temp is probably what
> you'll need.

Agreed.  The new version creates "darcs-record-X" where X is the next
available number, as generated from Lock.world_readable_temp.

The various functions that need this file now have it passed in as a
parameter.  The number of parameters for these functions is getting a
bit large.  I wonder if there is a better way?

Also, I'm not sure calling world_readable_temp directly is what we
want, since there could be a race condition.  Perhaps it would be
better to open the file handle as soon as we get it?  I attempted
this, but there are so many file-related wrappers in Darcs that I got
overwhelmed with how to go about this.

Finally, Ian suggested the file should be put in _darcs somewhere
instead.  Do others agree?

The current state of the patch is attached.

--
Zachary P. Landau <[EMAIL PROTECTED]>

Attachment: save_long_comment_file_if_a_test_fails_during_record_v2.dpatch
Description: Binary data

_______________________________________________
darcs-devel mailing list
[email protected]
http://www.abridgegame.org/cgi-bin/mailman/listinfo/darcs-devel

Reply via email to