Mocketry problems in 3.0
I recently checked the progress on TxText and I saw that half of the tests fail under 3.0 due to an incompatibility in Mocketry on BlockClosure>>#argumentNames trying to decompile the block contents. This feature is no longer available, instead you have to access the AST directly: BlockClosure >> #argumentNames ^ self sourceNode arguments collect: [ :each | each name ] Would it be possible to integrate that on a 3.0 branch? It's quite a pity to waste so many tests :)
On Oct 11, 2013, at 9:20 PM, Camillo Bruni <camillobruni@gmail.com> wrote:
I recently checked the progress on TxText and I saw that half of the tests fail under 3.0 due to an incompatibility in Mocketry on BlockClosure>>#argumentNames trying to decompile the block contents. This feature is no longer available, instead you have to access the AST directly:
BlockClosure >> #argumentNames ^ self sourceNode arguments collect: [ :each | each name ]
Would it be possible to integrate that on a 3.0 branch? It's quite a pity to waste so many tests :)
We can add #argumentNames on block in Pharo3⦠makes sense to have. Marcus
As Markus said it is good idea to add #argumentNames to block. What exactly you want to see at TxText? I ask because Igor make big progress on his design implementation which I think will replace my work totally. So if you want to join his development you don't need my code at all (remove it) But my version (which is at configuration) is usable and works nice for end user applications. I actually use it for my tasks. 2013/10/11 Camillo Bruni <camillobruni@gmail.com>
I recently checked the progress on TxText and I saw that half of the tests fail under 3.0 due to an incompatibility in Mocketry on BlockClosure>>#argumentNames trying to decompile the block contents. This feature is no longer available, instead you have to access the AST directly:
BlockClosure >> #argumentNames ^ self sourceNode arguments collect: [ :each | each name ]
Would it be possible to integrate that on a 3.0 branch? It's quite a pity to waste so many tests :)
On Oct 11, 2013, at 9:47 PM, Denis Kudriashov <dionisiydk@gmail.com> wrote:
As Markus said it is good idea to add #argumentNames to block.
What exactly you want to see at TxText? I ask because Igor make big progress on his design implementation which I think will replace my work totally. So if you want to join his development you don't need my code at all (remove it)
Not totally from what I saw and monday all the tests were green and we published the code because it was the end of a large refactoring to get immutable objects into place and having tests were cool. When we start a new behavior like copyText we started with tests. Stef
But my version (which is at configuration) is usable and works nice for end user applications. I actually use it for my tasks.
2013/10/11 Camillo Bruni <camillobruni@gmail.com> I recently checked the progress on TxText and I saw that half of the tests fail under 3.0 due to an incompatibility in Mocketry on BlockClosure>>#argumentNames trying to decompile the block contents. This feature is no longer available, instead you have to access the AST directly:
BlockClosure >> #argumentNames ^ self sourceNode arguments collect: [ :each | each name ]
Would it be possible to integrate that on a 3.0 branch? It's quite a pity to waste so many tests :)
When I look at code last time I see totally separated hierarchies of layout objects and morphs with prefix Athens. So I think my code is not planned to be used. Maybe we will separate our works at different packages. Also I almost finished my design ideas with mutable objects and active text model. At the end it will be interesting competition of our design decisions. 2013/10/12 Stéphane Ducasse <stephane.ducasse@inria.fr>
On Oct 11, 2013, at 9:47 PM, Denis Kudriashov <dionisiydk@gmail.com> wrote:
As Markus said it is good idea to add #argumentNames to block.
What exactly you want to see at TxText? I ask because Igor make big progress on his design implementation which I think will replace my work totally. So if you want to join his development you don't need my code at all (remove it)
Not totally from what I saw and monday all the tests were green and we published the code because it was the end of a large refactoring to get immutable objects into place and having tests were cool.
When we start a new behavior like copyText we started with tests.
Stef
But my version (which is at configuration) is usable and works nice for end user applications. I actually use it for my tasks.
2013/10/11 Camillo Bruni <camillobruni@gmail.com>
I recently checked the progress on TxText and I saw that half of the tests fail under 3.0 due to an incompatibility in Mocketry on BlockClosure>>#argumentNames trying to decompile the block contents. This feature is no longer available, instead you have to access the AST directly:
BlockClosure >> #argumentNames ^ self sourceNode arguments collect: [ :each | each name ]
Would it be possible to integrate that on a 3.0 branch? It's quite a pity to waste so many tests :)
On 2013-10-12, at 07:00, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
On Oct 11, 2013, at 9:47 PM, Denis Kudriashov <dionisiydk@gmail.com> wrote:
As Markus said it is good idea to add #argumentNames to block.
What exactly you want to see at TxText? I ask because Igor make big progress on his design implementation which I think will replace my work totally. So if you want to join his development you don't need my code at all (remove it)
Not totally from what I saw and monday all the tests were green and we published the code because it was the end of a large refactoring to get immutable objects into place and having tests were cool.
When we start a new behavior like copyText we started with tests.
I doubt that all the tests were ever green. half of the tests never ran in 3.0: https://ci.inria.fr/pharo-contribution/job/TxText/ Well, might just be a problem of the configuration, but that is hard to guess
As Markus said it is good idea to add #argumentNames to block.
What exactly you want to see at TxText? I ask because Igor make big progress on his design implementation which I think will replace my work totally. So if you want to join his development you don't need my code at all (remove it)
Not totally from what I saw and monday all the tests were green and we published the code because it was the end of a large refactoring to get immutable objects into place and having tests were cool.
When we start a new behavior like copyText we started with tests.
I doubt that all the tests were ever green.
the one of txModel package were because we worked until there were green. So I'm sure about that :)
half of the tests never ran in 3.0: https://ci.inria.fr/pharo-contribution/job/TxText/ Well, might just be a problem of the configuration, but that is hard to guess
participants (4)
-
Camillo Bruni -
Denis Kudriashov -
Marcus Denker -
Stéphane Ducasse