Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

> > > On 25/04/11 20:00, Johan Hake wrote: > > > Martin: > > > Why is it important to have the _count in the repr of the form? I guess > that > (Not sure if I got the quote author correct here.) The ufl Form does not have a _count... The _count of e.g. Coefficient objects is a critical property,

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

I'm not sure if this is safe or even solves anything. If there are circular references a.b = b; b.a = a, a.__del__ won't be called if the reference from b.a is still intact, since a.__del__ is called when the reference count of a goes to 0. Adding the __del__ function will also stop the cyclic refe

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 09:03, Martin Sandve Alnæs wrote: > See other mail. I don't see that it solves anything, it doesn't seem > related to anything I've read about in this thread, and it has a > potential backside in hindering the garbage collector in Python. I may > be wrong, but nobody has answered my o

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 09:56, Garth N. Wells wrote: > > > On 26/04/11 09:03, Martin Sandve Alnæs wrote: >> See other mail. I don't see that it solves anything, it doesn't seem >> related to anything I've read about in this thread, and it has a >> potential backside in hindering the garbage collector in Py

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26 April 2011 10:56, Garth N. Wells wrote: > > > On 26/04/11 09:03, Martin Sandve Alnæs wrote: > > See other mail. I don't see that it solves anything, it doesn't seem > > related to anything I've read about in this thread, and it has a > > potential backside in hindering the garbage collector

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26 April 2011 11:09, Garth N. Wells wrote: > > > On 26/04/11 09:56, Garth N. Wells wrote: > > > > > > On 26/04/11 09:03, Martin Sandve Alnæs wrote: > >> See other mail. I don't see that it solves anything, it doesn't seem > >> related to anything I've read about in this thread, and it has a >

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 12:25, Martin Sandve Alnæs wrote: > On 26 April 2011 11:09, Garth N. Wells > wrote: > > > > On 26/04/11 09:56, Garth N. Wells wrote: > > > > > > On 26/04/11 09:03, Martin Sandve Alnæs wrote: > >> See other mail. I don't see that it solve

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 01:22:28PM +0200, Martin Sandve Alnæs wrote: > On 26 April 2011 10:56, Garth N. Wells wrote: > > > > On 26/04/11 09:03, Martin Sandve Aln s wrote: > > See other mail. I don't see that it solves anything, it doesn't seem > > related to anything I've read about in

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 12:22, Martin Sandve Alnæs wrote: > On 26 April 2011 10:56, Garth N. Wells > wrote: > > > > On 26/04/11 09:03, Martin Sandve Alnæs wrote: > > See other mail. I don't see that it solves anything, it doesn't seem > > related to anything I've read

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 12:43:06PM +0100, Garth N. Wells wrote: > > > On 26/04/11 12:22, Martin Sandve Alnæs wrote: > > On 26 April 2011 10:56, Garth N. Wells > > wrote: > > > > > > > > On 26/04/11 09:03, Martin Sandve Alnæs wrote: > > > See other mail. I don't see

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

Caching is ok precicely _because_ the form is immutable. Martin On 26 April 2011 13:43, Garth N. Wells wrote: > > > On 26/04/11 12:22, Martin Sandve Alnæs wrote: > > On 26 April 2011 10:56, Garth N. Wells > > wrote: > > > > > > > > On 26/04/11 09:03, Martin Sandve A

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

So will you add it back? I can do it since it's my fault removing it, but I assume you want to handle it yourself. ;-) -- Anders On Tue, Apr 26, 2011 at 01:46:59PM +0200, Martin Sandve Alnæs wrote: > Caching is ok precicely _because_ the form is immutable. > > Martin > > On 26 April 2011 13:43,

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

Feel free to do it :) Martin On 26 April 2011 13:55, Anders Logg wrote: > So will you add it back? I can do it since it's my fault removing it, > but I assume you want to handle it yourself. ;-) > > -- > Anders > > > On Tue, Apr 26, 2011 at 01:46:59PM +0200, Martin Sandve Alnæs wrote: > > Cach

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

It feels good that you trust me enough to handle it. ;-) Will add it sometime this afternoon and then we can revisit the JIT compiler caching. -- Anders On Tue, Apr 26, 2011 at 01:57:40PM +0200, Martin Sandve Alnæs wrote: > Feel free to do it :) > > Martin > > > On 26 April 2011 13:55, Anders L

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 02:00:50PM +0200, Anders Logg wrote: > It feels good that you trust me enough to handle it. ;-) > > Will add it sometime this afternoon and then we can revisit the JIT > compiler caching. I'm getting confused here... Looking at preprocess.py in UFL, I see this: def prepro

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 13:51, Anders Logg wrote: > On Tue, Apr 26, 2011 at 02:00:50PM +0200, Anders Logg wrote: >> It feels good that you trust me enough to handle it. ;-) >> >> Will add it sometime this afternoon and then we can revisit the JIT >> compiler caching. > > I'm getting confused here... Looking

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 03:45:22PM +0100, Garth N. Wells wrote: > > > On 26/04/11 13:51, Anders Logg wrote: > > On Tue, Apr 26, 2011 at 02:00:50PM +0200, Anders Logg wrote: > >> It feels good that you trust me enough to handle it. ;-) > >> > >> Will add it sometime this afternoon and then we can re

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 00:01:04 Martin Sandve Alnæs wrote: > > > On 25/04/11 20:00, Johan Hake wrote: > > > > Martin: > > > > Why is it important to have the _count in the repr of the form? I > > > > guess > > > > that > > (Not sure if I got the quote author correct here.) > > The ufl Form doe

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 15:55, Anders Logg wrote: > On Tue, Apr 26, 2011 at 03:45:22PM +0100, Garth N. Wells wrote: >> >> >> On 26/04/11 13:51, Anders Logg wrote: >>> On Tue, Apr 26, 2011 at 02:00:50PM +0200, Anders Logg wrote: It feels good that you trust me enough to handle it. ;-) Will add

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 03:59:52PM +0100, Garth N. Wells wrote: > > > On 26/04/11 15:55, Anders Logg wrote: > > On Tue, Apr 26, 2011 at 03:45:22PM +0100, Garth N. Wells wrote: > >> > >> > >> On 26/04/11 13:51, Anders Logg wrote: > >>> On Tue, Apr 26, 2011 at 02:00:50PM +0200, Anders Logg wrote: > >

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 01:01:55 Martin Sandve Alnæs wrote: > I'm not sure if this is safe or even solves anything. > If there are circular references a.b = b; b.a = a, > a.__del__ won't be called if the reference from > b.a is still intact, since a.__del__ is called when > the reference count of

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 16:07, Anders Logg wrote: > On Tue, Apr 26, 2011 at 03:59:52PM +0100, Garth N. Wells wrote: >> >> >> On 26/04/11 15:55, Anders Logg wrote: >>> On Tue, Apr 26, 2011 at 03:45:22PM +0100, Garth N. Wells wrote: On 26/04/11 13:51, Anders Logg wrote: > On Tue, Apr 26, 201

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

Why can't we just go with the original design of Martin? In jit_compiler() we just call: form_data = form.form_data() This will generate the form data, including the preprocessed form attached to it, if the form_data is not already generated. Then we get the preprocessed form by: form_data

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:07:47 Anders Logg wrote: > On Tue, Apr 26, 2011 at 03:59:52PM +0100, Garth N. Wells wrote: > > On 26/04/11 15:55, Anders Logg wrote: > > > On Tue, Apr 26, 2011 at 03:45:22PM +0100, Garth N. Wells wrote: > > >> On 26/04/11 13:51, Anders Logg wrote: > > >>> On Tue, Apr 26,

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:16:29 Garth N. Wells wrote: > On 26/04/11 16:07, Anders Logg wrote: > > On Tue, Apr 26, 2011 at 03:59:52PM +0100, Garth N. Wells wrote: > >> On 26/04/11 15:55, Anders Logg wrote: > >>> On Tue, Apr 26, 2011 at 03:45:22PM +0100, Garth N. Wells wrote: > On 26/04/11 13:

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

Is no one listening??? Look at preprocess.py in UFL. It caches the preprocessed form and the preprocessed form holds the form data. This is the same as Martin's original design as far as I can tell. The difference is that a call to preprocess is necessary but that just returns the cached form. -

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 16:31, Johan Hake wrote: > On Tuesday April 26 2011 08:16:29 Garth N. Wells wrote: >> On 26/04/11 16:07, Anders Logg wrote: >>> On Tue, Apr 26, 2011 at 03:59:52PM +0100, Garth N. Wells wrote: On 26/04/11 15:55, Anders Logg wrote: > On Tue, Apr 26, 2011 at 03:45:22PM +0100, Ga

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: > On 26/04/11 16:31, Johan Hake wrote: > > On Tuesday April 26 2011 08:16:29 Garth N. Wells wrote: > >> On 26/04/11 16:07, Anders Logg wrote: > >>> On Tue, Apr 26, 2011 at 03:59:52PM +0100, Garth N. Wells wrote: > On 26/04/11 15:55, Ander

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 16:32, Anders Logg wrote: > Is no one listening??? > > Look at preprocess.py in UFL. It caches the preprocessed form and the > preprocessed form holds the form data. This is the same as Martin's > original design as far as I can tell. > Put some print statements and you'll see that

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: > On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: > > On 26/04/11 16:31, Johan Hake wrote: > > > On Tuesday April 26 2011 08:16:29 Garth N. Wells wrote: > > >> On 26/04/11 16:07, Anders Logg wrote: > > >>> On Tue, Apr 26, 2011 at 0

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:32:24 Anders Logg wrote: > Is no one listening??? I am. > Look at preprocess.py in UFL. It caches the preprocessed form and the > preprocessed form holds the form data. This is the same as Martin's > original design as far as I can tell. > > The difference is that a

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: > > On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: > > > On 26/04/11 16:31, Johan Hake wrote: > > > > On Tuesday April 26 2011 08:16:29 Garth N. Wells wrote: > > > >> On 26/04

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 16:39, Johan Hake wrote: > On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: >> On 26/04/11 16:31, Johan Hake wrote: >>> On Tuesday April 26 2011 08:16:29 Garth N. Wells wrote: On 26/04/11 16:07, Anders Logg wrote: > On Tue, Apr 26, 2011 at 03:59:52PM +0100, Garth N. We

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26 April 2011 17:08, Johan Hake wrote: > On Tuesday April 26 2011 01:01:55 Martin Sandve Alnæs wrote: > > I'm not sure if this is safe or even solves anything. > > If there are circular references a.b = b; b.a = a, > > a.__del__ won't be called if the reference from > > b.a is still intact, si

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 16:44, Johan Hake wrote: > On Tuesday April 26 2011 08:42:32 Anders Logg wrote: >> On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: >>> On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: On 26/04/11 16:31, Johan Hake wrote: > On Tuesday April 26 2011 08:16:29

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26 April 2011 17:47, Martin Sandve Alnæs wrote: > On 26 April 2011 17:08, Johan Hake wrote: > >> On Tuesday April 26 2011 01:01:55 Martin Sandve Alnæs wrote: >> > I'm not sure if this is safe or even solves anything. >> > If there are circular references a.b = b; b.a = a, >> > a.__del__ won't

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:46:26 Garth N. Wells wrote: > On 26/04/11 16:39, Johan Hake wrote: > > On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: > >> On 26/04/11 16:31, Johan Hake wrote: > >>> On Tuesday April 26 2011 08:16:29 Garth N. Wells wrote: > On 26/04/11 16:07, Anders Logg wr

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:50:31 Martin Sandve Alnæs wrote: > On 26 April 2011 17:47, Martin Sandve Alnæs wrote: > > On 26 April 2011 17:08, Johan Hake wrote: > >> On Tuesday April 26 2011 01:01:55 Martin Sandve Alnæs wrote: > >> > I'm not sure if this is safe or even solves anything. > >> > If

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 08:44:24AM -0700, Johan Hake wrote: > On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > > On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: > > > On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: > > > > On 26/04/11 16:31, Johan Hake wrote: > > > > > On

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 08:48:33 Garth N. Wells wrote: > On 26/04/11 16:44, Johan Hake wrote: > > On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > >> On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: > >>> On Tuesday April 26 2011 08:33:11 Garth N. Wells wrote: > On 26/04/11 1

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 09:01:37AM -0700, Johan Hake wrote: > On Tuesday April 26 2011 08:48:33 Garth N. Wells wrote: > > On 26/04/11 16:44, Johan Hake wrote: > > > On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > > >> On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: > > >>> On Tue

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

If you want to play around with weakref to understand it, try this snippet: import weakref class A: def __init__(s, n): s.n = n def __del__(s): print "d", s.n a = A("orig") b = A("prep") c = A("formdata") a.fd = c b.fd = c c.a = weakref.ref(a) c.b = weakref.ref(b) print "C" del c print "A" del a

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 09:03:55 Anders Logg wrote: > On Tue, Apr 26, 2011 at 09:01:37AM -0700, Johan Hake wrote: > > On Tuesday April 26 2011 08:48:33 Garth N. Wells wrote: > > > On 26/04/11 16:44, Johan Hake wrote: > > > > On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > > > >> On Tue, Apr

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26 April 2011 18:07, Johan Hake wrote: > On Tuesday April 26 2011 09:03:55 Anders Logg wrote: > > On Tue, Apr 26, 2011 at 09:01:37AM -0700, Johan Hake wrote: > > > On Tuesday April 26 2011 08:48:33 Garth N. Wells wrote: > > > > On 26/04/11 16:44, Johan Hake wrote: > > > > > On Tuesday April 26

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 17:03, Anders Logg wrote: > On Tue, Apr 26, 2011 at 09:01:37AM -0700, Johan Hake wrote: >> On Tuesday April 26 2011 08:48:33 Garth N. Wells wrote: >>> On 26/04/11 16:44, Johan Hake wrote: On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > On Tue, Apr 26, 2011 at 08:39:30A

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 09:01:35 Anders Logg wrote: > On Tue, Apr 26, 2011 at 08:44:24AM -0700, Johan Hake wrote: > > On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > > > On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: > > > > On Tuesday April 26 2011 08:33:11 Garth N. Wells wrot

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26 April 2011 18:10, Johan Hake wrote: > On Tuesday April 26 2011 09:01:35 Anders Logg wrote: > > On Tue, Apr 26, 2011 at 08:44:24AM -0700, Johan Hake wrote: > > > On Tuesday April 26 2011 08:42:32 Anders Logg wrote: > > > > On Tue, Apr 26, 2011 at 08:39:30AM -0700, Johan Hake wrote: > > > > >

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 06:12:26PM +0200, Martin Sandve Alnæs wrote: > On 26 April 2011 18:10, Johan Hake wrote: > > On Tuesday April 26 2011 09:01:35 Anders Logg wrote: > > On Tue, Apr 26, 2011 at 08:44:24AM -0700, Johan Hake wrote: > > > On Tuesday April 26 2011 08:42:32 Anders Logg

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On 26/04/11 17:16, Anders Logg wrote: > On Tue, Apr 26, 2011 at 06:12:26PM +0200, Martin Sandve Alnæs wrote: >> On 26 April 2011 18:10, Johan Hake wrote: >> >> On Tuesday April 26 2011 09:01:35 Anders Logg wrote: >> > On Tue, Apr 26, 2011 at 08:44:24AM -0700, Johan Hake wrote: >> > >

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tuesday April 26 2011 09:18:15 Garth N. Wells wrote: > On 26/04/11 17:16, Anders Logg wrote: > > On Tue, Apr 26, 2011 at 06:12:26PM +0200, Martin Sandve Alnæs wrote: > >> On 26 April 2011 18:10, Johan Hake wrote: > >> On Tuesday April 26 2011 09:01:35 Anders Logg wrote: > >> > On Tue, A

Re: [Ffc] [Ufl] [Bug 769811] [NEW] JIT cache problem with id(form)

On Tue, Apr 26, 2011 at 09:24:02AM -0700, Johan Hake wrote: > On Tuesday April 26 2011 09:18:15 Garth N. Wells wrote: > > On 26/04/11 17:16, Anders Logg wrote: > > > On Tue, Apr 26, 2011 at 06:12:26PM +0200, Martin Sandve Alnæs wrote: > > >> On 26 April 2011 18:10, Johan Hake wrote: > > >> On