[PyCUDA] Any interest in a PyCUDA stack exchange?
ely at seas.harvard.edu
Fri Mar 16 14:45:19 PDT 2012
My thinking was that it's a bit more cumbersome to look up archived PyCUDA
mailing list items, whether online or in my gmail search, to re-read them
for their applicability to my problem, and to parse the code discussed in
them. With a stackexchange, it's usually much easier to locate items that
are relevant from the past if I encounter them later, and since the code is
formatted with coloring and certain syntax niceties, just a lot easier to
read. Multiplied across dozens of mailing list entries that become relevant
to me, these extras go a long way. None of them are strictly necessary of
course, but why not have them when it's as easy as responding on
stackexchange instead of via email?
It is true that we can always just use stackoverflow. But the main point
was trying to get the very talented folks who answer things in this email
format to instead make cleaner, easier to read/find/index/search answers on
a stackexchange (which could just be stackoverflow). The way I see it is
there are two disparate answer communities, this mailing list vs. gpu
related tags on stackoverflow. I feel like it would be so much more useful
going forward if they were all just in one place, and stackoverflow seems
like the better place.
Since it doesn't seem like most folks care very much and that some people
actually prefer the email interface, perhaps I'll just make my own bookmark
website or something that attempts to index the pycuda mail digests by
their specific contents rather than just their titles or something.
On Thu, Mar 15, 2012 at 3:17 PM, Eli Stevens (Gmail)
<wickedgrey at gmail.com>wrote:
> Slightly off-topic: what's the benefit of having a dedicated stack
> exchange, vs. using Stack Overflow? I can only think of downsides
> related to fracturing the existing set of answers and responders, esp.
> for ones that aren't directly related to PyCUDA.
> On topic: I don't see a reason to not just use Stack Overflow.
> On Thu, Mar 15, 2012 at 11:31 AM, David Mertens
> <dcmertens.perl at gmail.com> wrote:
> > Being a Perl guy that mostly follows this list. You can probably count
> > number of Perl programmers that use CUDA or OpenCL in their scripts on
> > hand, but I happen to be one of them, and I would prefer a non-Python
> > centric topic, like plain CUDA, or CUDA/OpenCL. On such an exchange,
> > questions about PyCUDA would be fine, but keeping the forum open for
> > PerlCUDA, RubyCUDA, etc, would be appreciated, at least by me. :-)
> > On Wed, Mar 14, 2012 at 9:03 AM, Ely Spears <ely at seas.harvard.edu>
> >> I don't know if this has been brought up or discussed before; my
> >> if it has. It seems like a natural thing to do as more and more people
> >> developing with PyCUDA. It seems like it would be an easier place to
> >> questions, rather than reading the back-and-forths in email digests.
> >> Potentially it could even be a GPU stackexchange to open up questions
> >> regular CUDA and other paradigms. Thoughts? Immediate strong objections?
> >> If we want to do this, the right place to inquire is
> >> < http://area51.stackexchange.com/ >.
> >> --
> >> Ely M. Spears
> >> ---------------------
> >> http://people.seas.harvard.edu/~ely/
> >> Harvard GVI Group:
> >> http://gvi.seas.harvard.edu/
> >> gv at seas.harvard.edu
> >> Mailing list:
> >> https://lists.seas.harvard.edu/mailman/listinfo/gv
> >> _______________________________________________
> >> PyCUDA mailing list
> >> PyCUDA at tiker.net
> >> http://lists.tiker.net/listinfo/pycuda
> > --
> > "Debugging is twice as hard as writing the code in the first place.
> > Therefore, if you write the code as cleverly as possible, you are,
> > by definition, not smart enough to debug it." -- Brian Kernighan
> > _______________________________________________
> > PyCUDA mailing list
> > PyCUDA at tiker.net
> > http://lists.tiker.net/listinfo/pycuda
Ely M. Spears
Harvard GVI Group:
gv at seas.harvard.edu
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the PyCUDA