Hi, On Sat, Apr 30, 2011 at 5:21 AM, Ralf Gommers wrote:

On Wed, Apr 27, 2011 at 8:52 PM, Matthew Brett wrote:

...

Hi,

This is just to follow up on a dead thread of mine a little while back.

I was asking about letters for Clint Whaley's tenure case, from numpy, but I realized that I don't know who 'numpy' is :)

Is there in fact a numpy steering group?    Who is best to write letters representing the 'numpy community'?

At http://scipy.org/Developer_Zone there's a list of people under a big header "steering committee". It seems to me that writing such a letter representing the community is one of the purposes that committee could serve.

Ah - yes - thanks for the reply. In the interests of general transparency - and given that no-one from that group has replied to this email - how should the group best be addressed? By personal email? That seems to break the open-source matra of everything on-list: http://producingoss.com/en/setting-tone.html#avoid-private-discussions Thanks again, Matthew

Hi, On Tue, May 3, 2011 at 7:58 PM, Robert Kern wrote:

On Tue, May 3, 2011 at 12:07, Matthew Brett wrote:

...

Hi,

On Sat, Apr 30, 2011 at 5:21 AM, Ralf Gommers wrote:

...

On Wed, Apr 27, 2011 at 8:52 PM, Matthew Brett wrote:

...

Hi,

This is just to follow up on a dead thread of mine a little while back.

I was asking about letters for Clint Whaley's tenure case, from numpy, but I realized that I don't know who 'numpy' is :)

Is there in fact a numpy steering group?    Who is best to write letters representing the 'numpy community'?

At http://scipy.org/Developer_Zone there's a list of people under a big header "steering committee". It seems to me that writing such a letter representing the community is one of the purposes that committee could serve.

Ah - yes - thanks for the reply.

In the interests of general transparency - and given that no-one from that group has replied to this email - how should the group best be addressed?   By personal email?  That seems to break the open-source matra of everything on-list:

http://producingoss.com/en/setting-tone.html#avoid-private-discussions

Having project-relevant *discussions* on-list doesn't preclude getting someone's *attention* off-list.

Yes, that's true. My worry was that, having put the question on the list, and not had an answer, it might send a bad signal if it was obvious that I had only got a reply because I'd asked for one off-list.

I can't speak for the rest of the group, but as for myself, if you would like to draft such a letter, I'm sure I will agree with its contents.

Thank you - sadly I am not confident in deserving your confidence, but I will do my best to say something sensible. Any objections to a public google doc? See you, Matthew

Hi, On Wed, May 4, 2011 at 9:24 AM, Robert Kern wrote:

On Wed, May 4, 2011 at 11:14, Matthew Brett wrote:

...

Hi,

On Tue, May 3, 2011 at 7:58 PM, Robert Kern wrote:

...

...

I can't speak for the rest of the group, but as for myself, if you would like to draft such a letter, I'm sure I will agree with its contents.

Thank you - sadly I am not confident in deserving your confidence, but I will do my best to say something sensible.   Any objections to a public google doc?

Even better!

I've put up a draft here: numpy-whaley-support - https://docs.google.com/document/d/1gPhUUjWqNpRatw90kCqL1WPWvn1yicf2VAowWSyHlno/edit?hl=en_US&authkey=CPv49_cK I didn't know who to put as signatories. Maybe an extended steering group like (from http://scipy.org/Developer_Zone): Jarrod Millman Eric Jones Robert Kern Travis Oliphant Stefan van der Walt plus: Pauli Ralf Chuck or something like that? Anyone else care to sign / edit? Mark W for example? Sorry, I haven't been following the numpy commits very carefully of late. Best, Matthew

Hi, On Thu, May 26, 2011 at 9:32 PM, Pearu Peterson wrote:

Hi,

Would it be possible to setup a signing system where anyone who would like to support Clint could sign and advertise the system on relevant mailing lists? This would provide larger body of supporters for this letter and perhaps will have greater impact to whom the letter will be addressed. Personally, I would be happy to sign to such a letter.

On the letter: the letter should also mention scipy community as they benefit most from the ATLAS speed.

Maybe it would be best phrased then as 'numpy and scipy developers' instead of the steering group? I'm not sure how this kind of thing works for tenure letters, I would guess that if there are a very large number of signatures it might be difficult to see who is being represented... I'm open to suggestions. I can also ask Clint. I've added you as an editor - would you consider adding your name at the end, and maybe something about scipy? - you know the scipy blas / lapack stuff much better than I do. Cheers, Matthew

On Fri, May 27, 2011 at 7:39 AM, Matthew Brett wrote:

Hi,

...

Hi,

Would it be possible to setup a signing system where anyone who would

On Thu, May 26, 2011 at 9:32 PM, Pearu Peterson wrote: like

...

to support Clint could sign and advertise the system on relevant mailing lists? This would provide larger body of supporters for this letter and perhaps will have greater impact to whom the letter will be addressed. Personally, I would be happy to sign to such a letter.

On the letter: the letter should also mention scipy community as they benefit most from the ATLAS speed.

Maybe it would be best phrased then as 'numpy and scipy developers' instead of the steering group?

I'm not sure how this kind of thing works for tenure letters, I would guess that if there are a very large number of signatures it might be difficult to see who is being represented... I'm open to suggestions. I can also ask Clint.

I've added you as an editor - would you consider adding your name at the end, and maybe something about scipy? - you know the scipy blas / lapack stuff much better than I do.

Done for adding the name. The document is currently numpy oriented and I am not sure where to enter with scipy.. Technical summary of the situation with scipy blas/lapack stuff: The main difference in between numpy and scipy lapack-wise is that numpy has a lite C version of few lapack routines in case the lapack libraries are not available for building it while for scipy the lapack libraries are prerequisites as scipy provides interfaces to a much larger number of lapack routines. Having ATLAS in addition would greatly increase the performance of these routines. Pearu