[darcs-devel] darcs patch: Start converting FastPackedString into a... (and 4 more)

[David Roundy]

On Tue, Feb 05, 2008 at 02:34:57AM -0800, Eric Kow wrote:
> Second attempt at sending this.
> Note that this appears to pass all tests.  You have define HAVE_BYTESTRING
> by hand, though.

Applied, thanks! And I plan to write a configure test for this.

> Sun Feb  3 23:51:11 GMT 2008  Eric Kow <E.Y.Kow at brighton.ac.uk>
>   * Start converting FastPackedString into a Data.ByteString wrapper.
> 
> Tue Feb  5 09:06:59 GMT 2008  Eric Kow <E.Y.Kow at brighton.ac.uk>
>   * Add OldFastPackedString for people without the latest bytestring.
>   
>   Data.ByteString was included with GHC 6.6, but it seems the Data.ByteString.Internal
>   part of the API was only introduced in a later version.
> 
> Tue Feb  5 09:27:19 GMT 2008  Eric Kow <E.Y.Kow at brighton.ac.uk>
>   * Restore old FastPackedString version of linesPS and unlinesPS.
>   They behave slightly differently with respect to newlines.
> 
> Tue Feb  5 09:28:29 GMT 2008  Eric Kow <E.Y.Kow at brighton.ac.uk>
>   * Restore some original FastPackedString code.
>   It was hanging on some tests, or running out of stack space.

I'm puzzled by this one.  Have you been able to figure out which bytestring
function was ill-behaving? I'd be interested to know, and I suspect dons
would be too.

> Tue Feb  5 09:30:00 GMT 2008  Eric Kow <E.Y.Kow at brighton.ac.uk>
>   * Stricter use of BI.toForeignPtr in FastPackedString.
>   
>   I don't know if this actually makes a difference.  Indeed, the intention behind
>   this patch is to avoid introducing any extra laziness where none was found before.