12 Jul
2007
12 Jul
'07
6:15 p.m.
Where exactly do you get a recursion?
It looks like that this is exactly the point in time where I added
code to avoid recursion when dealing with recursive descriptions. I
guess the method causing trouble is
MAValidatorVisitor>>#visitDescription:?
Lukas
On 12.07.2007, at 17:48, Rick Flower wrote:
Lukas,
I dropped a note to Thomas Koschate, who did the first ports of
Magritte
over to VW, asking about moving the current version of Magritte
over to VW
in the hopes of having something to use w/ Seaside 2.7 or 2.8a
(instead of
2.6).. Anyway, he indicated that he had tried porting over
Magritte-all-lr.224 last month but ran into an issue w/ infinite
recursion
as part of some perceived exception handling changes.. Anyway, do
you know
off the top of your head if anything changed between versions 192/193
(from Feb 16th, 2007) and 224 in how exception handling is dealt
with in
Magritte? I've not had a chance to look into it yet after his latest
comments early this morning (I don't have VW/Squeak at work). Anyway,
just thought I'd ask.. Thanks!
-- Rick
--
Lukas Renggli
http://www.lukas-renggli.ch
12 Jul
12 Jul
6:29 p.m.
New subject: Differences w/ exception handling in Magritte..?
On Thu, July 12, 2007 9:15 am, Lukas Renggli wrote:
Where exactly do you get a recursion?
It looks like that this is exactly the point in time where I added
code to avoid recursion when dealing with recursive descriptions. I
guess the method causing trouble is
MAValidatorVisitor>>#visitDescription:?
I'm not sure exactly -- I'll have to load it to find out -- I'll get back
to you with some more details.. Thanks for the heads-up though -- at least
it's a place to start looking.
6:35 p.m.
New subject: Differences w/ exception handling in Magritte..?
On 7/12/07, Lukas Renggli <renggli(a)iam.unibe.ch> wrote:
Where exactly do you get a recursion?
It looks like that this is exactly the point in time where I added
code to avoid recursion when dealing with recursive descriptions. I
guess the method causing trouble is
MAValidatorVisitor>>#visitDescription:?
That's exactly the spot. The #resignalAs: is causing VW some heartburn.
--
============================================================
Thomas Koschate
6:44 p.m.
New subject: Differences w/ exception handling in Magritte..?
That's exactly the spot. The #resignalAs: is
causing VW some
heartburn.
I don't exactly remember why I copy the exception there. That seems
rather strange and presumably causes VW to end up in an infinite
recursion.
Please try with the following code that works -- as far as I can see
right now -- well in Squeak:
visitDescription: aDescription
"Validate the current object using aDescription within an exception
handler to avoid running further tests that might cause error-cascades."
aDescription isVisible ifFalse: [ ^ self ].
[ self validate: self object using: aDescription ]
on: MAValidationError
do: [ :err | err beResumable; pass ]
Lukas
--
Lukas Renggli
http://www.lukas-renggli.ch
8:42 p.m.
New subject: Differences w/ exception handling in Magritte..?
On 7/12/07, Lukas Renggli <renggli(a)iam.unibe.ch> wrote:
Please try with the following code that works -- as far as I can see
right now -- well in Squeak:
visitDescription: aDescription
That solves the problem nicely for that one. However, we run into the same
issue with MAStringReader>>#visitElementDescription:. Morever, there's
actually a bug with this method, since #resignalAs: is expecting an instance
of an exception, and you're passing in a class.
While I remember, a request to help with future ports: Please avoid the use
of {} to define arrays, as VisualWorks doesn't know how to deal with this.
--
============================================================
Thomas Koschate
9:10 p.m.
New subject: Differences w/ exception handling in Magritte..?
That solves the problem nicely for that one. However,
we run into
the same issue with MAStringReader>>#visitElementDescription:.
Morever, there's actually a bug with this method, since
#resignalAs: is expecting an instance of an exception, and you're
passing in a class.
In Squeak a class and an instance works. I fixed that to create a new
instance instead.
While I remember, a request to help with future ports:
Please
avoid the use of {} to define arrays, as VisualWorks doesn't know
how to deal with this.
I found one occurrence in a test. Are there others?
The problems are fixed:
Name: Magritte-All-lr.227
Author: lr
Time: 12 July 2007, 9:08:32 pm
UUID: eadc01b1-8663-4f91-8acc-c67a30f133d4
Ancestors: Magritte-All-lr.226
Dependencies: Magritte-Model-lr.272, Magritte-Tests-lr.116, Magritte-
Seaside-lr.232, Magritte-Morph-lr.38
- for platform compatibility:
- create an instance for #resignalAl:
- removed single use of {} in test
--
Lukas Renggli
http://www.lukas-renggli.ch
9:23 p.m.
New subject: Differences w/ exception handling in Magritte..?
On 7/12/07, Lukas Renggli <renggli(a)iam.unibe.ch> wrote:
In Squeak a class and an instance works. I fixed that to create a new
No, it was only the one occurrence.
Thanks for the quick turnaround!
--
============================================================
Thomas Koschate +1 416.716.5949
koschate(a)omasko.com FAX: +1 905.451.9268
instance instead.
I'm sorry, I may not have made myself understood. The infinite recursion
error occurs as a result of the #resignalAs:, even if an instance is
correctly supplied.
I found one occurrence in a test. Are there others?
9:51 p.m.
New subject: Differences w/ exception handling in Magritte..?
Ok, no more senders of #resignalAl: in
Name: Magritte-All-lr.228
Author: lr
Time: 12 July 2007, 9:50:42 pm
UUID: d7bdc877-4341-4986-8a7d-810edd165050
Ancestors: Magritte-All-lr.227
Dependencies: Magritte-Model-lr.273, Magritte-Tests-lr.116, Magritte-
Seaside-lr.232, Magritte-Morph-lr.38
- avoid another infinite look with #resignalAs:
On 12.07.2007, at 21:23, Thomas Koschate wrote:
On 7/12/07, Lukas Renggli <renggli(a)iam.unibe.ch> wrote:
In Squeak a class and an instance works. I fixed that to create a new
instance instead.
I'm sorry, I may not have made myself understood. The infinite
recursion error occurs as a result of the #resignalAs:, even if an
instance is correctly supplied.
I found one occurrence in a test. Are there others?
No, it was only the one occurrence.
Thanks for the quick turnaround!
--
============================================================
Thomas Koschate +1 416.716.5949
koschate(a)omasko.com FAX: +1 905.451.9268
--
Lukas Renggli
http://www.lukas-renggli.ch
10:29 p.m.
New subject: Differences w/ exception handling in Magritte..?
On 7/12/07, Lukas Renggli <renggli(a)iam.unibe.ch> wrote:
Ok, no more senders of #resignalAl: in
We're getting close now. I just need to work through a few more Squeak/VW
dialect differences and we'll be up to date again.
--
============================================================
Thomas Koschate +1 416.716.5949
koschate(a)omasko.com FAX: +1 905.451.9268
7:23 p.m.
New subject: Differences w/ exception handling in Magritte..?
Name: Magritte-All-lr.226
Author: lr
Time: 12 July 2007, 7:22:28 pm
UUID: 72307372-3019-412b-94df-ab0524468fca
Ancestors: Magritte-All-lr.225
Dependencies: Magritte-Model-lr.271, Magritte-Tests-lr.115, Magritte-
Seaside-lr.232, Magritte-Morph-lr.38
- trying to improve the way resumeable errors are thrown, so that it
works on VW as well
- added tests to exactly check the way resumeable errors are handled
On 12.07.2007, at 18:35, Thomas Koschate wrote:
On 7/12/07, Lukas Renggli <renggli(a)iam.unibe.ch> wrote: Where
exactly do you get a recursion?
It looks like that this is exactly the point in time where I added
code to avoid recursion when dealing with recursive descriptions. I
guess the method causing trouble is
MAValidatorVisitor>>#visitDescription:?
That's exactly the spot. The #resignalAs: is causing VW some
heartburn.
--
============================================================
Thomas Koschate
_______________________________________________
SmallWiki, Magritte, Pier and Related Tools ...
https://www.iam.unibe.ch/mailman/listinfo/smallwiki
--
Lukas Renggli
http://www.lukas-renggli.ch
6132
days inactive
6132
days old
9 comments
3 participants
participants (3)
-
Lukas Renggli -
Rick Flower -
Thomas Koschate