Claus Reinke wrote:
Could you please send patches using either 'darcs send', or diff -c/-u. Ordinary diffs aren't particularly readable, and don't apply with patch, at least not without typing in the filenames manually (unless I'm missing something?).

that may be because of the path names used by 'darcs diff', or because
the line numbers no longer match, as this was the third patch in a series.

i have had bad experiences with 'darcs send', which requires me to do
'darcs record': i never know when, whether, or in what form my patches
make it into the main repo, which means that my repo is out of sync
after a record (not to mention that my changes no longer show up in
whatsnew, fooling me into believing i've got a pristine tree, or that darcs add so much context to even the smallest patches).

anyway, i've recorded/sent my three ghci patches and attach the
result for your convenience. i'll do a 'darcs unrecord' locally before
the next pull, perhaps that avoids my troubles from previous times.

Thanks, that makes it a lot easier to review the changes.

  - Overall, I like the additions, especially :show languages/packages

  - I wonder whether :set should show *all* the dynamic flags, rather than
    just the GHCi-related ones?  Or perhaps just the flags that have
    changed from their defaults?

  - Move ghciFlags lower down, maybe as a local function to setCmd

  - Any chance you could refactor runCommands to be a bit more readable?
    I find this style quite incomprehensible:

    maybe (return ())
        (((bool (return ()) (runCommands getCmd)) =<<)
          . ghciHandle handler . doCommand) mb_cmd

    eh? :-)

Cheers,
        Simon

_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to