Hi,
the API of Pragma was changed in quite dangerous way. In past we had: - #selector to access to method selector - #keyword to access to pragma selector now we have: - #methodSelector to access to method selector - #selector to access to pragma selector That means that the meaning of the message #selector was changed and it can cause some problems (like some currently failing tests [1]). The #selector message was deprecated in the middle of the transition process so most of usages should be already rewritten to #methodSelector but we need to be aware of it. [1] https://pharo.fogbugz.com/f/cases/19333/17-more-tests-failing-due-to-MockSettings-class-Object-doesNotUnderstand-mocksystemsettings-SettingTreeBuilder-mocksystemse Cheers -- Pavel |
Le 11/11/2016 à 10:55, Pavel Krivanek a écrit :
> Hi, > > the API of Pragma was changed in quite dangerous way. > In past we had: > - #selector to access to method selector > - #keyword to access to pragma selector > > now we have: > - #methodSelector to access to method selector > - #selector to access to pragma selector > > That means that the meaning of the message #selector was changed and it > can cause some problems (like some currently failing tests [1]). The > #selector message was deprecated in the middle of the transition process > so most of usages should be already rewritten to #methodSelector but we > need to be aware of it. > > [1] https://pharo.fogbugz.com/f/cases/19333/17-more-tests-failing-due-to-MockSettings-class-Object-doesNotUnderstand-mocksystemsettings-SettingTreeBuilder-mocksystemse > > Cheers > -- Pavel Thank you for the report. This broke the latest Metacello. I opened an issue: https://github.com/dalehenrich/metacello-work/issues/424 -- Cyril Ferlicot http://www.synectique.eu 2 rue Jacques Prévert 01, 59650 Villeneuve d'ascq France signature.asc (836 bytes) Download Attachment |
In reply to this post by Pavel Krivanek-3
On 11/11/16 10:55, Pavel Krivanek wrote:
> Hi, > > the API of Pragma was changed in quite dangerous way. > In past we had: > - #selector to access to method selector > - #keyword to access to pragma selector > > now we have: > - #methodSelector to access to method selector > - #selector to access to pragma selector Revert and first do a deprecation phase Stephan |
On 11/11/2016 15:58, Stephan Eggermont wrote:
> Revert and first do a deprecation phase > > Stephan > > > +1 It would be cool to go by a deprecation phase indeed. -- Cyril Ferlicot http://www.synectique.eu 2 rue Jacques Prévert 01, 59650 Villeneuve d'ascq France signature.asc (817 bytes) Download Attachment |
In reply to this post by Pavel Krivanek-3
On 11/11/16 10:55, Pavel Krivanek wrote:
> That means that the meaning of the message #selector was changed and it > can cause some problems (like some currently failing tests [1]). The > #selector message was deprecated in the middle of the transition process > so most of usages should be already rewritten to #methodSelector but we > need to be aware of it. As long as we do not systematically collect selectors sent from our sources we cannot know that most users are rewritten. That might be a valuable addition to our development process. For some experiments, take a look at DeprecationFinder Stephan |
In reply to this post by Pavel Krivanek-3
2016-11-11 10:55 GMT+01:00 Pavel Krivanek <[hidden email]>:
|
2016-11-13 10:06 GMT+01:00 Nicolai Hess <[hidden email]>:
And FileList can not show .cs files anymore. I think both has something to do with a lock (?) during the transform of the autodeprecation.
|
Well, we will revert the change -- Pavel 2016-11-13 17:33 GMT+01:00 Nicolai Hess <[hidden email]>:
|
2016-11-14 9:42 GMT+01:00 Pavel Krivanek <[hidden email]>:
revert the autodeprecation or the new api ? I made a fix for case 19336 Finder and FileList can now work with the new api.
|
Revert the new API. It was mistake to integrate it in the code freeze phase. -- Pavel 2016-11-14 9:53 GMT+01:00 Nicolai Hess <[hidden email]>:
|
yep, that’s accurate :) Esteban
|
Great. Le 14 nov. 2016 11:23, "Esteban Lorenzano" <[hidden email]> a écrit :
|
In reply to this post by Pavel Krivanek-3
Hi,
Please do not revert the API. Just comment out the deprecation. The code in Moose was already updated to the new API and I presume other projects might be in the same situation. Cheers, Doru > On Nov 14, 2016, at 10:00 AM, Pavel Krivanek <[hidden email]> wrote: > > Revert the new API. It was mistake to integrate it in the code freeze phase. > > -- Pavel > > 2016-11-14 9:53 GMT+01:00 Nicolai Hess <[hidden email]>: > > > 2016-11-14 9:42 GMT+01:00 Pavel Krivanek <[hidden email]>: > Well, we will revert the change > > > revert the autodeprecation or the new api ? > I made a fix for case 19336 Finder and FileList can now work with the new api. > > -- Pavel > > 2016-11-13 17:33 GMT+01:00 Nicolai Hess <[hidden email]>: > > > 2016-11-13 10:06 GMT+01:00 Nicolai Hess <[hidden email]>: > > > 2016-11-11 10:55 GMT+01:00 Pavel Krivanek <[hidden email]>: > Hi, > > the API of Pragma was changed in quite dangerous way. > In past we had: > - #selector to access to method selector > - #keyword to access to pragma selector > > now we have: > - #methodSelector to access to method selector > - #selector to access to pragma selector > > That means that the meaning of the message #selector was changed and it can cause some problems (like some currently failing tests [1]). The #selector message was deprecated in the middle of the transition process so most of usages should be already rewritten to #methodSelector but we need to be aware of it. > > [1] https://pharo.fogbugz.com/f/cases/19333/17-more-tests-failing-due-to-MockSettings-class-Object-doesNotUnderstand-mocksystemsettings-SettingTreeBuilder-mocksystemse > > And it broke finder. > 19336 Finder broken: Can not search for pragmas > > And FileList can not show .cs files anymore. > I think both has something to do with a lock (?) during the transform of the autodeprecation. > > > > Cheers > -- Pavel > > > > > -- www.tudorgirba.com www.feenk.com "Be rather willing to give than demanding to get." |
On 14/11/2016 13:44, Tudor Girba wrote:
> Hi, > > Please do not revert the API. Just comment out the deprecation. The code in Moose was already updated to the new API and I presume other projects might be in the same situation. > > Cheers, > Doru > > > -- > www.tudorgirba.com > www.feenk.com > > "Be rather willing to give than demanding to get." > > > breaking change without warning is something companies can fear. This is even more important when there is such a dangerous change. #selector is still here but do something different… This is bad to just change it in one go. Maybe some projects updated but most companies do not follow the development version of Pharo. -- Cyril Ferlicot http://www.synectique.eu 2 rue Jacques Prévert 01, 59650 Villeneuve d'ascq France signature.asc (817 bytes) Download Attachment |
Hi,
Indeed, I forgot that #selector is doing something else. This is a bit of a mess. Either we revert it today, or we leave it. Otherwise, more projects will adopt the new API only to break again afterwards. Doru > On Nov 14, 2016, at 1:50 PM, Cyril Ferlicot D. <[hidden email]> wrote: > > On 14/11/2016 13:44, Tudor Girba wrote: >> Hi, >> >> Please do not revert the API. Just comment out the deprecation. The code in Moose was already updated to the new API and I presume other projects might be in the same situation. >> >> Cheers, >> Doru >> >> >> -- >> www.tudorgirba.com >> www.feenk.com >> >> "Be rather willing to give than demanding to get." >> >> >> > > Changing the API without a deprecation phase is bad. This kind of > breaking change without warning is something companies can fear. This is > even more important when there is such a dangerous change. #selector is > still here but do something different… This is bad to just change it in > one go. > > Maybe some projects updated but most companies do not follow the > development version of Pharo. > > -- > Cyril Ferlicot > > http://www.synectique.eu > > 2 rue Jacques Prévert 01, > 59650 Villeneuve d'ascq France -- www.tudorgirba.com www.feenk.com "It's not how it is, it is how we see it." |
we need to revert.
this is a huge mess in a very inappropriate moment :) Esteban > On 14 Nov 2016, at 13:56, Tudor Girba <[hidden email]> wrote: > > Hi, > > Indeed, I forgot that #selector is doing something else. This is a bit of a mess. Either we revert it today, or we leave it. Otherwise, more projects will adopt the new API only to break again afterwards. > > Doru > >> On Nov 14, 2016, at 1:50 PM, Cyril Ferlicot D. <[hidden email]> wrote: >> >> On 14/11/2016 13:44, Tudor Girba wrote: >>> Hi, >>> >>> Please do not revert the API. Just comment out the deprecation. The code in Moose was already updated to the new API and I presume other projects might be in the same situation. >>> >>> Cheers, >>> Doru >>> >>> >>> -- >>> www.tudorgirba.com >>> www.feenk.com >>> >>> "Be rather willing to give than demanding to get." >>> >>> >>> >> >> Changing the API without a deprecation phase is bad. This kind of >> breaking change without warning is something companies can fear. This is >> even more important when there is such a dangerous change. #selector is >> still here but do something different… This is bad to just change it in >> one go. >> >> Maybe some projects updated but most companies do not follow the >> development version of Pharo. >> >> -- >> Cyril Ferlicot >> >> http://www.synectique.eu >> >> 2 rue Jacques Prévert 01, >> 59650 Villeneuve d'ascq France > > -- > www.tudorgirba.com > www.feenk.com > > "It's not how it is, it is how we see it." > > |
Agreed. So who is doing it?
Doru > On Nov 14, 2016, at 2:16 PM, Esteban Lorenzano <[hidden email]> wrote: > > we need to revert. > this is a huge mess in a very inappropriate moment :) > > Esteban > >> On 14 Nov 2016, at 13:56, Tudor Girba <[hidden email]> wrote: >> >> Hi, >> >> Indeed, I forgot that #selector is doing something else. This is a bit of a mess. Either we revert it today, or we leave it. Otherwise, more projects will adopt the new API only to break again afterwards. >> >> Doru >> >>> On Nov 14, 2016, at 1:50 PM, Cyril Ferlicot D. <[hidden email]> wrote: >>> >>> On 14/11/2016 13:44, Tudor Girba wrote: >>>> Hi, >>>> >>>> Please do not revert the API. Just comment out the deprecation. The code in Moose was already updated to the new API and I presume other projects might be in the same situation. >>>> >>>> Cheers, >>>> Doru >>>> >>>> >>>> -- >>>> www.tudorgirba.com >>>> www.feenk.com >>>> >>>> "Be rather willing to give than demanding to get." >>>> >>>> >>>> >>> >>> Changing the API without a deprecation phase is bad. This kind of >>> breaking change without warning is something companies can fear. This is >>> even more important when there is such a dangerous change. #selector is >>> still here but do something different… This is bad to just change it in >>> one go. >>> >>> Maybe some projects updated but most companies do not follow the >>> development version of Pharo. >>> >>> -- >>> Cyril Ferlicot >>> >>> http://www.synectique.eu >>> >>> 2 rue Jacques Prévert 01, >>> 59650 Villeneuve d'ascq France >> >> -- >> www.tudorgirba.com >> www.feenk.com >> >> "It's not how it is, it is how we see it." >> >> > > -- www.tudorgirba.com www.feenk.com "Don't give to get. Just give." |
We should not revert whole update so I created an issue: -- Pavel 2016-11-14 14:30 GMT+01:00 Tudor Girba <[hidden email]>: Agreed. So who is doing it? |
In reply to this post by EstebanLM
Breaking Metacello in Pharo 6: Not good. Revert: +1 On Mon, Nov 14, 2016 at 2:16 PM, Esteban Lorenzano <[hidden email]> wrote: we need to revert. |
2016-11-14 11:27 GMT-03:00 [hidden email] <[hidden email]>:
> Breaking Metacello in Pharo 6: Not good. Revert: +1 Breaking Metacello in any version = Not good. Metacello is very backward compatible but breaking pragma semantics would require a rewrite of many configurations that work okay today and would require a new version of Metacello that contemplates the new selectors based on some criteria. Regards! Esteban A. Maringolo |
Free forum by Nabble | Edit this page |