bug#20507: 25.0.50 eieio and packages depending on it

bug#20507: 25.0.50 eieio and packages depending on it

[Pierre Lorenzon]

Hi,

I already posted two bug reports 20467 and 20491 but each time
I correct something I get a new error. Last time it was an
error in srecode dictionaries.

Anyway I suspect that eieio implementation has been change
without doing necessary modification in codes that use
eieio. surely it is very time consuming and people who are
responsible for that ahd no time to do it. Surely it is
absolutley normal for a development version not to be fully
functional. Anyway I wonder why my bug report 20647 where I
proposed patches to correct some errors got no answer. Time
problem probably.

But if you need help for the eieio stuff please simply tell
me. I think I am able to contribute to that. But please tell me
more precisely which are the goals so that I do not make
something that is not in the spec ! Or simply tell me if
there's a good place to know what has to be done with the eieio
project and all its dependencies.


Regards

Pierre

bug#20507: 25.0.50 eieio and packages depending on it

[Stefan Monnier]

> I already posted two bug reports 20467 and 20491 but each time
> I correct something I get a new error.

I'm sorry about that.  When I installed those changes in EIEIO there were
a few bugs reported, but since then, it's been fairly quiet, so I was
under the impression that things are mostly OK.

> Anyway I suspect that eieio implementation has been change
> without doing necessary modification in codes that use
> eieio.

I did test the new code against CEDET's use, but in only fairly limited
ways (mostly, enabling semantic-mode and using it to perform completion
on a struct field name in regex.c).

> Anyway I wonder why my bug report 20647 where I proposed patches to
> correct some errors got no answer.  Time problem probably.

Yup.  Have you had a chance to try my reply in bug#20491?

> But if you need help for the eieio stuff please simply tell me.

Please do send more of those bug reports and (especially) patches.
Recipes to reproduce those problems are also very welcome (test cases
to add to test/automated are even better), since in many cases I have no
idea how the code is supposed to be used.

> I think I am able to contribute to that. But please tell me
> more precisely which are the goals so that I do not make
> something that is not in the spec ! Or simply tell me if
> there's a good place to know what has to be done with the eieio
> project and all its dependencies.

Mostly what needs to be done is to make sure that old code still works,
and when it doesn't decide whether it should be fixed by changing EIEIO
to be more bug-compatible or by changing the old code.


        Stefan

bug#20507: 25.0.50 eieio and packages depending on it

[Eric Ludlam]

On 05/05/2015 02:56 PM, Stefan Monnier wrote:
>
>> >I think I am able to contribute to that. But please tell me
>> >more precisely which are the goals so that I do not make
>> >something that is not in the spec ! Or simply tell me if
>> >there's a good place to know what has to be done with the eieio
>> >project and all its dependencies.
> Mostly what needs to be done is to make sure that old code still works,
> and when it doesn't decide whether it should be fixed by changing EIEIO
> to be more bug-compatible or by changing the old code.
>

Hi Stefan,

CEDET has an extensive test suite, but most of it hasn't been merged 
into Emacs' tests.  I suspect the easiest way for you to get some good 
coverage is to pull CEDET from the sourceforge repository, build it 
using your development version of Emacs, and then do:

make utest-batch

and when that works

make itest-batch


The unit tests were failing for me with Emacs 25 a couple months back 
and I fixed a bunch of issues to make it work with some of the planned 
changes to eieio.  I'm not sure where it stands today as I haven't had 
time to mess with it recently.

The tests include all the tools that use eieio, and exercise a wide 
range of features from Semantic's completion to the SRecode insertion 
macros.  If you pass these tests, I'd have high confidence your eieio 
changes are in good shape.

The unit tests usually only take a minute or so, but the interactive 
tests can take several minutes.  The tests require you have many 
external tools installed which I assume you do, such as make, automake, 
gcc, and a few other things.

Eric

bug#20507: 25.0.50 eieio and packages depending on it

[Pierre Lorenzon]

Hi Stefan

Thanks for your answer.


From: Stefan Monnier <monnier@iro.umontreal.ca>
Subject: Re: bug#20507: 25.0.50 eieio and packages depending on it
Date: Tue, 05 May 2015 14:56:56 -0400

>> I already posted two bug reports 20467 and 20491 but each time
>> I correct something I get a new error.
> 
> I'm sorry about that.  When I installed those changes in EIEIO there were
> a few bugs reported, but since then, it's been fairly quiet, so I was
> under the impression that things are mostly OK.

  Unfortunately it is maybe not exactly the case and there is
  still things to improve. Anyway I cannot say how much there's
  to do. In fact I discover the bugs each one after another.



> 
>> Anyway I suspect that eieio implementation has been change
>> without doing necessary modification in codes that use
>> eieio.
> 
> I did test the new code against CEDET's use, but in only fairly limited
> ways (mostly, enabling semantic-mode and using it to perform completion
> on a struct field name in regex.c).

  OK That's why bugs in ede implementation for instance have
  not been detected. Anyway I'll continu to track them !





> 
>> Anyway I wonder why my bug report 20647 where I proposed patches to
>> correct some errors got no answer.  Time problem probably.
> 
> Yup.  Have you had a chance to try my reply in bug#20491?

  Yes I did the modification to the code that you suggested and
  it fixed the bug at this place.





> 
>> But if you need help for the eieio stuff please simply tell me.
> 
> Please do send more of those bug reports and (especially) patches.
> Recipes to reproduce those problems are also very welcome (test cases
> to add to test/automated are even better), since in many cases I have no
> idea how the code is supposed to be used.

  No problem I'll try to make the stuff work and send patches.





> 
>> I think I am able to contribute to that. But please tell me
>> more precisely which are the goals so that I do not make
>> something that is not in the spec ! Or simply tell me if
>> there's a good place to know what has to be done with the eieio
>> project and all its dependencies.
> 
> Mostly what needs to be done is to make sure that old code still works,
> and when it doesn't decide whether it should be fixed by changing EIEIO
> to be more bug-compatible or by changing the old code.



  OK 


> 
> 
>         Stefan

	  Thanks again

	  Pierre

bug#20507: 25.0.50 eieio and packages depending on it

[Stefan Monnier]

> CEDET has an extensive test suite, but most of it hasn't been merged into
> Emacs' tests.  I suspect the easiest way for you to get some good coverage
> is to pull CEDET from the sourceforge repository, build it using your
> development version of Emacs, and then do:

Thanks, that sounds very helpful.


        Stefan

bug#20507: 25.0.50 eieio and packages depending on it

[npostavs]

close 20507 
quit

Closing since there's no actual bug reported here.  I think this would
have been better as an emacs-devel thread.