[Gsll-devel] inverse of cl-array

[Liam Healy]

OK the latest commit in master now has the following feature:

#'make-marray will optionally take a :cl-array argument.  This should
be a CL array generated by #'make-ffa, with the same element type as
specified in the first argument to make-marray.  The symbol 'make-ffa
is exported for users.

Be cautious here.  There are things you can do that could mess things
up - changing the contents of the CL array from outside the marray
object, sending the same CL array to two different marrays, etc.

Liam



On Tue, Feb 10, 2009 at 10:12 PM, Liam Healy <lhealy at common-lisp.net> wrote:
> There isn't an inverse to cl-array at present.
>
> Some thoughts about this, without having yet dived into the code:  my
> first recommendation is to just make-marray for everything by default.
>  When you need a regular CL array, use cl-array to extract it, and
> then if you need to, copy it (see below).   The problem with supplying
> an array directly to be made into a marray is that it may not be
> "foreign friendly".  This won't make any difference on most
> implementations, but on SBCL, you will get unpredictable results
> because SBCL will avoid copying the array over to the C side and just
> send a pointer to the GSL functions.  I could perhaps add the ability
> to supply a foreign-friendly array, as made by make-ffa, as an
> :initial-contents for make-marray.
>
> If you are willing to always have the array copied, that could be done
> pretty easily too.   As a side point, please note that changing an
> element of an array extracted with cl-array will change the marray in
> SBCL and have unpredictable results on other implementations.  This is
> why I recommend you copy the array after you extract it.
>
> Liam
>
> On Tue, Feb 10, 2009 at 8:49 AM, Mirko Vukovic <mirko.vukovic at gmail.com> wrote:
>>
>>
>> On Tue, Feb 10, 2009 at 5:09 AM, Niitsuma Hirotaka
>> <hirotaka.niitsuma at gmail.com> wrote:
>>>
>>> I can not find convert function
>>> #A(...) -> #m(...)
>>>
>>> ( cl-array is conversion #m -> #a )
>>>
>>> Thus I write the following ,but that looks slow.
>>> --------------
>>> (defun coerce-array-to-list (in-array)
>>>   (map 'list
>>>           #'identity
>>>           (make-array (array-total-size in-array)
>>>                             :element-type (array-element-type in-array)
>>>                             :displaced-to in-array)))
>>> ;borrow from
>>> http://groups.google.com/group/comp.lang.lisp/browse_thread/thread/a925ca60d88d4047/cc06e579dfe840a7?lnk=gst&q=displaced+array
>>>
>>>
>>> (defun appropriate-name (element-type ar)
>>>  (make-marray element-type  :dimensions (array-dimensions ar)
>>> :initial-contents  (coerce-array-to-list ar))
>>> )
>>> ---------------
>>>
>>> usage:
>>> ( appropriate-name 'double-float #2A((1.0d0 2.0d0) (2.0d0 5.0d0)))
>>> ->
>>> #m((1.0d0 2.0d0) (2.0d0 5.0d0)))
>>>
>>> _______________________________________________
>>> Gsll-devel mailing list
>>> Gsll-devel at common-lisp.net
>>> http://common-lisp.net/cgi-bin/mailman/listinfo/gsll-devel
>>
>> I don't know an answer better than yours.  I would name the function
>> gsll-array.
>> Also, since marray's are objects, you may want to look into its constructor
>> to see how the array is stored.
>>
>> I was bitten a few times by the make-marray syntax that requires a list
>> input.
>>
>> (As far as I see, lisp's coerce syntax prevents us from customizing it to do
>> the job we want.)
>>
>> Mirko
>>
>> _______________________________________________
>> Gsll-devel mailing list
>> Gsll-devel at common-lisp.net
>> http://common-lisp.net/cgi-bin/mailman/listinfo/gsll-devel
>>
>>
>