[Haskell-cafe] Google SoC: Space profiling reloaded

[Patai Gergely]

Hello all,

I'm one of the lucky individuals accepted for this year's Summer of
Code. Here's the abstract of my proposal:

http://socghop.appspot.com/student_project/show/google/gsoc2009/haskell/t124022468245

There's less than a month left before I'm supposed to jump into coding,
and I'd love to hear about any little idea you think would make this
project even better! I created a project page with a rough draft of what
I intend to concentrate on, but the issue tracker is obviously begging
for content, so feel encouraged to populate it. :)

http://code.google.com/p/hp2any/

I also intend to expand the description later.

Gergely

-- 
http://www.fastmail.fm - The professional email service

___
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Haskell-cafe] Google SoC: Space profiling reloaded

[Claus Reinke]

http://socghop.appspot.com/student_project/show/google/gsoc2009/haskell/t124022468245

There's less than a month left before I'm supposed to jump into coding,
and I'd love to hear about any little idea you think would make this
project even better! I created a project page with a rough draft of what
I intend to concentrate on, but the issue tracker is obviously begging
for content, so feel encouraged to populate it. :)

http://code.google.com/p/hp2any/


Hi there,

I'm sure lots of Haskellers are looking forward to profiling improvements!-)

The wxHaskell applications page

http://wxhaskell.sourceforge.net/applications.html

mentions 


http://dready.org/projects/HPView/

which seems relevant.

Shortening the song and dance needed to get profiles would be nice
(eliminating the PS viewer dependency in the process).
So would not having to select profile type in advance (and not having
to repeat the profile runs again and again, with only one view each time).

I assume you're only going to look into visualization, not semantics?-)
http://www.haskell.org/pipermail/cvs-ghc/2008-October/045981.html

Also, there is a lot of space profiling support, but I'd sometimes like
more detailed time profiling (not just end-of-run summaries, but time
sliced profiling - in this time slice, most of the time was spent in x,
in the next time slice, most of the time was spent in y, these functions
are frequented in the beginning, but are not used towards the end, etc). 
Perhaps the recent support for code coverage could be bent to support 
something like this without too much hacking?


Keep us informed!
Claus


___
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe