[GSLL-devel] Bug or feature: grid-map and complex type destinations

[Mirko Vukovic]

FWIW, I downloaded the last gsd patch of Nov. 22, and the asd systems
load cleanly.  No more infinite loops.

My hypothesis was that the soft links in the gsd directory of grid.asd
and foreign-array.asd to their location
in the gsd/grid and gsd/foreign-array was the culprit.  But I tested
that on a separate tree, and asdf2 worked
fine.

I tried downloading and installing the version of Nov. 21, and this
time it loaded up cleanly too.   So, it is not
something in the files.




On Fri, Nov 19, 2010 at 10:12 PM, Mirko Vukovic <mirko.vukovic at gmail.com> wrote:
> No, the file was local.   The only remotely relevant item is that the
> file timestamps were those from the git repository (for example, if I
> download a version a few days old, the file timestamps will be from
> November 8 say).
>
> The following is from hazy (possibly incorrect memory): I was able to
> install grid by cd-in into gsd/grid and doing asdf from there.  Then
> when I tried to install foreign-array from gsd I got asdf looping
> again.  I have ASDF2 searching these directories recursively (as a
> tree).  (As I said, hazy/incorrect recollection as I was in a hurry to
> head home).  I will verify that next week.
>
> I am using ASDF2
>
> Mirko
>
> On Fri, Nov 19, 2010 at 8:53 PM, Liam Healy <lhealy at common-lisp.net> wrote:
>> That one I recognize.  By any chance do you have your files on a
>> server (like NFS)?  If so check to make sure the clocks are
>> synchronized.  If the asd file has a time in the future, ASDF goes
>> into an infinite loop.  We had an NFS server that had a significant
>> clock drift, and I would get bitten by that a lot.  I posted a bug to
>> the ASDF list but no one responded.
>>
>> Liam
>>
>> On Fri, Nov 19, 2010 at 6:04 PM, Mirko Vukovic <mirko.vukovic at gmail.com> wrote:
>>> thanks.
>>>
>>> Unfortunately, I cannot try it right now.  I am having some weird
>>> issues with asdf -- it enters into an infinite loop and exceeds the
>>> nesting limit.
>>>
>>> I will work on this next week.
>>>
>>> Mirko
>>>
>>>
>>>
>>> On Fri, Nov 19, 2010 at 5:21 PM, Liam Healy <lhealy at common-lisp.net> wrote:
>>>> Complex types are specified as (complex double-float), (complex
>>>> single-float), etc.  These work:
>>>>
>>>> (map-grid :source #1m(-1d0 -2d0 -3d0)
>>>>          :element-function #'sqrt
>>>>          :destination-specification '((foreign-array 3) (complex double-float)))
>>>> #2m(#C(0.0d0 1.0d0) #C(0.0d0 1.4142135623730951d0)
>>>>    #C(0.0d0 1.7320508075688772d0))
>>>>
>>>> (make-instance 'vector-complex-double-float :dimensions '(3)
>>>> :element-type '(complex double-float))
>>>> #2m(#C(0.0d0 0.0d0) #C(0.0d0 0.0d0) #C(0.0d0 0.0d0))
>>>>
>>>> Note: you need to pull the latest version of GSD for the first case to
>>>> work, 307ba9efba fixes a bug in using the type specification.
>>>>
>>>> Liam
>>>>
>>>>
>>>> On Fri, Nov 19, 2010 at 4:04 PM, Mirko Vukovic <mirko.vukovic at gmail.com> wrote:
>>>>> Hi,
>>>>>
>>>>> I am reporting this in case it is a gsd bug (it must be a bug
>>>>> somewhere: in my head, my setup, or gsd)
>>>>>
>>>>> Mapping the `sqrt' function over a vector of negative numbers as follows:
>>>>>
>>>>> grid> (map-grid :source #1m(-1d0 -2d0 -3d0)
>>>>>                :element-function #'sqrt
>>>>>                :destination-specification '((foreign-array 3) complex-double-float))
>>>>>
>>>>> gives an error:
>>>>>
>>>>> Element type COMPLEX-DOUBLE-FLOAT not permitted.
>>>>>   [Condition of type SIMPLE-ERROR]
>>>>>
>>>>> I traced it to a `make-instance' statement.
>>>>>
>>>>> I also tried to instantiate a vector-complex-double-float object
>>>>>
>>>>> grid> (make-instance 'vector-complex-double-float
>>>>>                     :dimensions '(3)
>>>>>                     :element-type 'complex-double-float)
>>>>>
>>>>> that resulted in the same error message.
>>>>>
>>>>> Mirko
>>>>>
>>>>> _______________________________________________
>>>>> GSLL-devel mailing list
>>>>> GSLL-devel at common-lisp.net
>>>>> http://common-lisp.net/cgi-bin/mailman/listinfo/gsll-devel
>>>>>
>>>>
>>>
>>
>