On Sat, 8 Nov 2008, Ganesh Sittampalam wrote:

> On Sat, 8 Nov 2008, Salvatore Insalaco wrote:
>
>> I believe that this change makes strictness semantics more explicit, and
>> has no performance penalties. If Don doesn't tell us anyway, I think
>> that it's better to keep it with B.readFile.
>
> The current data pipeline is
>
> file =[lazy]=> decompress =[lazy]=> strictify
>
> Your change is to
>
> file =[strict]=> decompress =[lazy]=> strictify
>
> I don't think that's a good idea in general, because the entire file now
> needs to be read into memory before decompression, whereas before it would
> stream nicely. As you say, Don (or Duncan) are best placed to comment
> further.

I've spoken to Duncan about this and he confirms what I said; using a 
strict bytestring will mean that the compressed file will all need to be 
in memory simultaneously, whereas previously it wouldn't.

So I'm strongly in favour of conditional compilation here.

Ganesh
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to