recover lost changes

To whom can answer me

Working with Pharo 3.0 I noticed for a while that I could not recover
class definitions (do it's) from the "Recent Changes".
I also lost class comments on file-out / file-in .

For the second problem I made the following test:
1. take a fresh Pharo 3.0 image (bundle for Mac OSX from Aug 29, 2014)
2. create MyClass with a class method and a comment.
3. file-out MyClass
4. remove MyClass
         on remove of MyClass I get an exception:
https://www.dropbox.com/s/unq89t84tq7r60b/removeClass.tiff?dl=0
5. file-in MyClass
The class is there - no error

But: With the same exception in other images the comment is missing
after file-in.

Is this known?
What does the exception mean?

--
Hartmut Krasemann

Königsberger Str. 41 c

D 22869 Schenefeld

Tel. 040.8307097

Mobil 0171.6451283

[hidden email]

Excerpts from Hartmut's message of 2015-02-10 15:58:13 +0100:
> Working with Pharo 3.0 I noticed for a while that I could not recover
> class definitions (do it's) from the "Recent Changes".

i noticed this too.
methods worked fine, but i couldn't recover methods to new classes until i
recreated the class manually.

greetings, martin.

--
eKita                   -   the online platform for your entire academic life
--
chief engineer                                                       eKita.co
pike programmer      pike.lysator.liu.se    caudium.net     societyserver.org
secretary                                                      beijinglug.org
mentor                                                           fossasia.org
foresight developer  foresightlinux.org                            realss.com
unix sysadmin
Martin Bähr          working in china        http://societyserver.org/mbaehr/

Yes this should be fixed.

Le 10/2/15 16:42, Martin Bähr a écrit :
> Excerpts from Hartmut's message of 2015-02-10 15:58:13 +0100:
>> Working with Pharo 3.0 I noticed for a while that I could not recover
>> class definitions (do it's) from the "Recent Changes".
> i noticed this too.
> methods worked fine, but i couldn't recover methods to new classes until i
> recreated the class manually.
>
> greetings, martin.
>

> Am 10.02.2015 um 20:59 schrieb stepharo <[hidden email]>:
>
> Yes this should be fixed.
>
You mean fix the current implementation or swap it for epicea?

Norbert

Norbert wrote:
>> Am 10.02.2015 um 20:59 schrieb stepharo <[hidden email]>:
>>
>> Yes this should be fixed.
>>
> You mean fix the current implementation or swap it for epicea?

The current implementation. It is a significant barrier to entry for
new smalltalkers and gives a very bad impression.
The epicea swap is a high-impact change, that is likely to
affect stability for some time.

Stephan

Yes but resources are still sparse. I'm usually arguing the same way like you. But at the moment I have the feeling we postpone the integration of too much too long. If the release is postponed until july I'd vote for integrating it now and iron it out. Otherwise it might be never integrated and that is our poor history of good projects dying because.

my 2 cents,

Norbert

Norbert wrote:
>Yes but resources are still sparse. I'm usually arguing the same way like you.
>But at the moment I have the feeling we postpone the integration of too much too long.
>If the release is postponed until july I'd vote for integrating it now and iron it out.
>Otherwise it might be never integrated and that is our poor history of
>good projects dying because.

I don't know the current state of epicea. To me, it sounds like something
that should have some time in real use before the code freeze. I agree that
a postponed release would make that time available if we can integrate
real soon now.

Stephan

Stephan Eggermont wrote

The current implementation

+1. It is really bad, especially for new users, that we've let it go on this long. It should have been fixed and backported to 3.0.

Marcus Denker-4 wrote

The issue is here

And the solution is right there in the comment!

Pharo Issue Tracker wrote

ChangesLog default
is not registered for the SystemAnnouncer.
Reinitializing it should help:
ChangesLog reset;default.

Works on both 3.0 and 4.0. I made a parent issue 14916 and an issue for 4.0 14918 (13290 was about 3.0).

Hartmut wrote

I also lost class comments on file-out / file-in .

I can not reproduce this in the lastest update (#30862). What image update version do you have? You can find out via WorldMenu->System->About

Also, when you say "

I remember at such case, missing comments after file in. I think we fixed it in 4.0 and back ported it too.

Am 11.02.2015 18:18 schrieb "Nicolai Hess" <[hidden email]>:
>
> I remember at such case, missing comments after file in. I think we fixed it in 4.0 and back ported it too.
>

14377

FileOut creates invalid comment entries for classes with class side methods

And 14405 for 3.0

> Am 11.02.2015 17:47 schrieb "Sean P. DeNigris" <[hidden email]>:
>
>> Hartmut wrote
>> > I also lost class comments on file-out / file-in .
>>
>> I can not reproduce this in the lastest update (#30862). What image update
>> version do you have? You can find out via WorldMenu->System->About
>>
>> Also, when you say "
>>
>>
>>
>> -----
>> Cheers,
>> Sean
>> --
>> View this message in context: http://forum.world.st/recover-lost-changes-tp4804929p4805162.html
>> Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com.
>>

Sean P. DeNigris wrote

I can not reproduce this...

Okay, I see. It only happens when Nautilus (the browser) is on the class-side. Looking deeper, there is in general a flaw in Nautilus - at least in announcement handling - because it doesn't take class vs. metaclass into account. I started fixing it, but don't have time now.

The basics for anyone that's interested: selected class could be a class, metaclass, or nil, so there needs to be a nil guard first, then something like selectedClass theNonMetaClass = aClass. Browsing (scoped to Nautilus) senders of #selectedClass is a good start. HTH

thanks sean.
I need to get back hacking in Pharo for real.

Stef

Le 11/2/15 18:58, Sean P. DeNigris a écrit :

Le 11/2/15 09:33, Stephan Eggermont a écrit :
> Norbert wrote:
>>> Am 10.02.2015 um 20:59 schrieb stepharo <[hidden email]>:
>>>
>>> Yes this should be fixed.
>>>
>> You mean fix the current implementation or swap it for epicea?
> The current implementation.

Yes

>   It is a significant barrier to entry for
> new smalltalkers and gives a very bad impression.
> The epicea swap is a high-impact change, that is likely to
> affect stability for some time.

No it is stable. We postponed it because martin was doing override in MC
instead of adding notification to MC
and registering to them. It will be fixed.

Stef
>
> Stephan
>
>

Yes but resources are still sparse. I'm usually arguing the same way
like you. But at the moment I have the feeling we postpone the
integration of too much too long. If the release is postponed until july
I'd vote for integrating it now and iron it out. Otherwise it might be
never integrated and that is our poor history of good projects dying
because. my 2 cents, Norbert

+1

Stef

Thanks to all of you ! 2. I can identify (and eventually remove) the empty class comments
written by the meta class

Hartmut

--
Hartmut Krasemann

Königsberger Str. 41 c

D 22869 Schenefeld

Tel. 040.8307097

Mobil 0171.6451283

[hidden email]