![](https://secure.gravatar.com/avatar/4d021a1d1319f36ad861ebef0eb5ba44.jpg?s=120&d=mm&r=g)
I just wanted to point people to the online version of the PEP. I'm still looking for comments and suggestions. The current version is here: http://projects.scipy.org/scipy/numpy/browser/trunk/numpy/doc/pep_buffer.txt -Travis
![](https://secure.gravatar.com/avatar/9ad321a1a9b37df1f7c401b602f98740.jpg?s=120&d=mm&r=g)
Travis Oliphant wrote:
Hi Travis: I'd like to help, but am not really prepared to comment at this time. It does look good at first glance. I need to review it in greater detail along with the current buffer code. Maybe this weekend. Do you have any kind of target dates that we need to keep in mind for this project? # Steve
![](https://secure.gravatar.com/avatar/9ad321a1a9b37df1f7c401b602f98740.jpg?s=120&d=mm&r=g)
Travis Oliphant wrote:
Hi Travis: Have you contacted PIL developers to see if they will support this? Your proposal looks like it would take a manageable amount of work to implement and shouldn't cause any long term maintenance issues. Have you had any response from the Python core developers? # Steve
![](https://secure.gravatar.com/avatar/3f692386259303b90d1967ad662a9eb0.jpg?s=120&d=mm&r=g)
Impressive work and an very intresting approach. I have yet to figure out pros/cons or the variety of options this will give when writing applications/modules. Big plus for giving ctypes a way out of their crappy struct definitions. A few quick questions: Is the concept here that items always are contiguous in memory (=why the strides cant be defined in the format-spec) ? What will PyObject_SizeFromFormat(''(5,5)5t") return ? What are the total memory size of format:'5t' shape:ndim=2,shape=(5,5),strides=NULL ? Huw would the "getbufferproc" function signature look inside the "X{}"-format spec ? //Torgil On 2/28/07, Travis Oliphant <oliphant@ee.byu.edu> wrote:
![](https://secure.gravatar.com/avatar/9ad321a1a9b37df1f7c401b602f98740.jpg?s=120&d=mm&r=g)
Travis Oliphant wrote:
Adding some use cases to the rationale might help convince the Python developers that this change is worth the development and maintenance burden. What about Python database adapters or data acquisition software? As the developers are very concerned with keeping their maintenance load manageable, a reference implementation would be persuasive, and they may want a commitment to help update affected standard library modules. # Steve
![](https://secure.gravatar.com/avatar/9ad321a1a9b37df1f7c401b602f98740.jpg?s=120&d=mm&r=g)
Travis Oliphant wrote:
Hi Travis: I'd like to help, but am not really prepared to comment at this time. It does look good at first glance. I need to review it in greater detail along with the current buffer code. Maybe this weekend. Do you have any kind of target dates that we need to keep in mind for this project? # Steve
![](https://secure.gravatar.com/avatar/9ad321a1a9b37df1f7c401b602f98740.jpg?s=120&d=mm&r=g)
Travis Oliphant wrote:
Hi Travis: Have you contacted PIL developers to see if they will support this? Your proposal looks like it would take a manageable amount of work to implement and shouldn't cause any long term maintenance issues. Have you had any response from the Python core developers? # Steve
![](https://secure.gravatar.com/avatar/3f692386259303b90d1967ad662a9eb0.jpg?s=120&d=mm&r=g)
Impressive work and an very intresting approach. I have yet to figure out pros/cons or the variety of options this will give when writing applications/modules. Big plus for giving ctypes a way out of their crappy struct definitions. A few quick questions: Is the concept here that items always are contiguous in memory (=why the strides cant be defined in the format-spec) ? What will PyObject_SizeFromFormat(''(5,5)5t") return ? What are the total memory size of format:'5t' shape:ndim=2,shape=(5,5),strides=NULL ? Huw would the "getbufferproc" function signature look inside the "X{}"-format spec ? //Torgil On 2/28/07, Travis Oliphant <oliphant@ee.byu.edu> wrote:
![](https://secure.gravatar.com/avatar/9ad321a1a9b37df1f7c401b602f98740.jpg?s=120&d=mm&r=g)
Travis Oliphant wrote:
Adding some use cases to the rationale might help convince the Python developers that this change is worth the development and maintenance burden. What about Python database adapters or data acquisition software? As the developers are very concerned with keeping their maintenance load manageable, a reference implementation would be persuasive, and they may want a commitment to help update affected standard library modules. # Steve
participants (3)
-
Steven H. Rogers
-
Torgil Svensson
-
Travis Oliphant