[darcs-users] darcs patch: Make Haskeline a required dependency (+ 4 more)

Judah Jacobson judah.jacobson at gmail.com
Tue Feb 3 18:33:49 UTC 2009


On Tue, Feb 3, 2009 at 10:01 AM, Petr Rockai <me at mornfall.net> wrote:
> Hi,
>
> I guess Salvatore could cross-review, I'm just going to read through this and
> if I encounter anything, note it inline.
>
>
> Don't set stdin's buffering; Haskeline will do that for us.
> -----------------------------------------------------------
> Removes without_buffering from a lot of places, as advertised. I'd like to hear
> a comment from Judah on how the buffering issue is handled by haskeline,
> though. My pet peeve with that keeping a key pressed to get past many patches
> screws up some internal state right now. It might be because the buffering is
> disabled all the time, but the program is not ready to read. IIUIC, haskeline
> will only disable buffering while it is ready to process input? Maybe that will
> actually fix that bug.

Yes, Haskeline temporarily disables buffering while reading user input
(i.e., only during a call to getInputLine or getInputChar).  So I
don't expect this to be an issue, but let me know if you encounter any
strange behavior.

> Don't use the bytestring-mmap package on Windows.
> -------------------------------------------------
> I would have expected cabal to deal with unavailability of bytestring-mmap on
> windows gracefully by itself? Does it cause any actual trouble this way? (Can't
> check as I don't have a windows host available.)

Unfortunately, when "cabal install" tries fetching dependencies from
hackage it isn't smart enough to avoid building the unix package on
Windows.  (Just doing a "runghc Setup configure/build/install"
probably works, though.)  Incidentally, darcs.cabal already does this
same trick for the terminfo package.

-Judah


More information about the darcs-users mailing list