[Pharo-project] Very strange behavior with MC

Stéphane Ducasse stephane.ducasse at inria.fr
Fri Jan 9 11:07:53 EST 2009


On Jan 9, 2009, at 4:26 PM, Adrian Lienhard wrote:

> No I didn't.
>
> But maybe we better recreate an image from the last version that
> didn't have this setting? I don't think it was introduced by a change
> because in my images that I updated myself the setting was not  
> changed.

Ok so may be we should redo a new core starting from the version you  
mentioned was
not having such behavior. Do you do it?

Stef

>
>
> Adrian
>
> On Jan 9, 2009, at 16:15 , Stéphane Ducasse wrote:
>
>> adrian
>>
>> did you use the cleaning expression of alex in your release?
>>
>> Stef
>>
>> On Jan 7, 2009, at 3:08 PM, Alexandre Bergel wrote:
>>
>>> Dear List,
>>>
>>> I dug in the Monticello melting-pot to find where the strange
>>> behavior
>>> some of you are encountering comes from.
>>>
>>> **The problem**
>>> The bug come from the "diffy versions". MC tries to save in
>>> PharoInbox
>>> a "diffy" version of your package. As far as I understood, a "diffy"
>>> version is a version of your package that contains just a delta from
>>> an ancestor. In theory this looks nice, but for some reasons it
>>> behaves incorrectly (the zip file is not probably created, it cannot
>>> be decompressed)
>>>
>>> **How to solve it**
>>> Recently (since 10183), the http repository for PharoInbox is set to
>>> store version in a diffy mode. You can identity http repository that
>>> are set that way:
>>> 	MCHttpRepository allInstances select: [:mc | mc alwaysStoreDiffs]
>>>
>>> The problem I mentioned in my previous email disappear symply by
>>> doing
>>> 	(MCHttpRepository allInstances select: [:mc | mc alwaysStoreDiffs])
>>> 		do: [:mc | mc doNotAlwaysStoreDiffs]
>>>
>>> After having executed the last snippet, storing packages in
>>> PharoInbox
>>> will not appear empty.
>>>
>>> What is left to be done, is why this diffy mechanism does not work  
>>> as
>>> expected. But I feel that we should not use it anyway since the
>>> ancestor link is not alway kept and is incorrectly handled (e.g.,
>>> cross-repository ancestors) and MC2 is now in the pipeline.
>>>
>>> Cheers,
>>> Alexandre
>>>
>>>
>>> On 6 Jan 2009, at 15:26, Alexandre Bergel wrote:
>>>
>>>> Dear all,
>>>>
>>>> I just save a new version in PharoInbox, but this version appears  
>>>> to
>>>> be empty. I checked on squeaksource [*].
>>>> In Pharo, it seems that two files are being saved (.143(142).mcd
>>>> and .143.mcz) [**].
>>>> when I browse .143(142).mcd I see all the content (probably because
>>>> it is local to my machine).
>>>> When I browse .143.mcz it is empty.
>>>>
>>>> What this .mcd file is about?
>>>> I am not the only one to experience this. Anyone had an idea what's
>>>> going on?
>>>> I haven't closely followed the list of changes that were included  
>>>> in
>>>> Pharo.
>>>>
>>>> Alexandre
>>>>
>>>> [*] http://bergel.eu/EmptyVersion.png
>>>> [**] http://bergel.eu/TwoSavedVersions.png
>>>> -- 
>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
>>>> Alexandre Bergel  http://www.bergel.eu
>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>> -- 
>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
>>> Alexandre Bergel  http://www.bergel.eu
>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>>>
>>>
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Pharo-project mailing list
>>> Pharo-project at lists.gforge.inria.fr
>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>>
>>
>>
>> _______________________________________________
>> Pharo-project mailing list
>> Pharo-project at lists.gforge.inria.fr
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
>
> _______________________________________________
> Pharo-project mailing list
> Pharo-project at lists.gforge.inria.fr
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>





More information about the Pharo-dev mailing list