[V3dot10] Re: Monticello Mysteries

[Keith Hodges]

The TTFFix loaded for me when I did it manually from a changeset. when I 
tried to make it part of a larger script it would reliably hang.

I loaded the fix and saved the package in which it occurs, 
'Multilingual'. Then used that in the larger script and that was ok.

This prompted me to allow changesets to be loaded in Installer using the 
Monticello MCPackageLoader which does atomic loading of methods. I will 
see if this helps the original case.

I am pondering the idea of modifying monticello to manage changesets anyway!

best regards

Keith

> --- "Edgar J. De Cleene" <edgardec2001 at yahoo.com.ar>
> wrote:
>
>   
>> Jerome:
>> I see you are using the notes and adding valuable
>> info.
>> How Ralph agree on solve problematic cs.
>> That only works for the TTFSpeed
>>
>>     
>>>> http://bugs.squeak.org/view.php?id=6348 is a
>>>>         
>> "worst case" for
>>     
>>>> Monticello, and I think that Monticello will
>>>>         
>> never handle cases like
>>     
>>>> this.
>>>>
>>>> I think the solution is to
>>>>   1) make MCZ files for the changes.  The MCZ
>>>>         
>> files will not load, but
>>     
>>>> will show the results of loading the .cs files
>>>>         
>
>
> I love it.  The roses will look red after they have
> been painted red.
>
> This bug implies a test.
>
> Are there MC files you can save but not load?
> I.e. Take a fresh image 
> make changes
> save mc package. (we will assume just one will be
> involved.)
>
> quit w/o saving. 
> Restart with the fresh image. 
> ask to load the just saved package and problems?
>
> For what class of changes would the above be true?
>
> TTF fonts depending on what is displayed when the
> changes are loaded?
>
> The weak array fix?
>
> It would be good to have a crash test or two for the
> doubters.
>
>
>
>   
>>>>   2) make an update that first loads the .cs
>>>>         
>> files and then loads the MCZ
>>     
>>>> files.
>>>>
>>>> Loading the MCZ files after loading the .cs files
>>>>         
>> should have no
>>     
>>>> effect on the methods that are in the image. 
>>>>         
>> However, it will keep
>>     
>>>> the image consistent from the point of view of
>>>>         
>> Monticello.
>>     
>
>
> Does that create a chicken and egg problem with the
> code repository.
>
> I.e. inorder to load code whose ancestor is xxx you
> have to start with an image that has already loaded
> xxx in it . Inother words any image pre-xxx will not
> update w/ mc alone?
>
>   
>>>> Probably the .cs files could be loaded in a
>>>>         
>> script that is part of the
>>     
>>>> MCZ files but I think that is not a good idea
>>>>         
>> because most people do
>>     
>>>> not expect a MCZ file to have a complex script.
>>>>
>>>> Do you think this will work?
>>>>
>>>> Are there other Mantis issues that have .cs files
>>>>         
>> that are too complex
>>     
>>>> for Monticello?    It would be good to make a
>>>>         
>> list of them on the
>>     
>>>> wiki.
>>>>
>>>> -Ralph
>>>>         
>>> TTFSpeedUp-dgd.1.cs into a fresh 7067 (no mc
>>>       
>> involved)
>> Blows for me, Keith said he could load without
>> trouble
>>     
> What is the critical difference between your
> experience and ours?
> What change makes it not work for you but work for us?
>
>
> This worked on my iMac (running classic vm):
> In fresh 7067 
> get a new morphic project and enter it.
> get a file list.
> point to TTFSpeedUp-dgd.1.cs and install.
> (it recompiles things but it works.)
> point to HelloWorld.text (a copy of dgd's test text.)
> grab a transcript and resize it so it is out of the
> way.
> Select and doit the HelloWorld example (it works 
> transcript says 1541 ms to run on my 400hz iMac.)
> So I had no problems loading the cs that way on my
> iMac.
>
> I tried it from a 7092 and had no problems either.
>
> I can't test the mc version of the load I don't get
> the .mcd to download from the 310 repository.
>
> Have no idea what causes that to be a problem.
>
>
>
>   
>>> will that work? What about a fresh 7092?
>>>       
>> All times what I test this loads and works.
>> Using Diego test (see Mantis) , redraw is 10 x
>> faster
>>
>>     
> These are quick thoughts. I pass them along to see if
> they stir up other ideas.
>
> I've replied rather quickly, so please forgive me if
> the thoughts have come out choppy.
>
> Yours in curiosity and service, --Jerome Peace
>
>
>  
> ____________________________________________________________________________________
> TV dinner still cooling? 
> Check out "Tonight's Picks" on Yahoo! TV.
> http://tv.yahoo.com/
> _______________________________________________
> V3dot10 mailing list
> V3dot10 at lists.squeakfoundation.org
> http://lists.squeakfoundation.org/mailman/listinfo/v3dot10
>
>