[Cuis] [ANN] Cuis-SecureSocket

[Juan Vuletich]

David Graham wrote:
> On 5/25/12 4:46 PM, Daniel Lyons wrote:
>> On Fri, May 25, 2012 at 04:40:17PM -0500, David Graham wrote:
>>> This is closely related to something that I'm genuinely curious
>>> about.  I don't consider myself experienced enough to fully
>>> understand all the issues surrounding smalltalk DVCS, but it seems
>>> like everyone is moving towards file-based DVCSs (i.e. git) for
>>> smalltalk code.  Is the future direction to move away from smalltalk
>>> tools and integrate file based tools?  Again, it may be my
>>> inexperience, but it seems like a strange trend for such a powerful
>>> programming environment.

I don't think there is a general trend. Cuis is not abandoning 
Monticello, Cuis never adopted it. I think Monticello falls into several 
of the reasons for not integrating it into Cuis:
- It is huge
- It is complex
- It is unstable
- Almost nobody understands it
- It is slow
- It is not really essential. You can code without it.

There is another question besides what goes in the image and what goes 
as optional code, and it is "What should be done in Smalltalk (either 
internal or external package) and what should be better left outside 
Smalltalk (to other apps, the OS, etc)?"

To me, the answer to this question has to do with what are my interests. 
In the Dynabook vision, Smalltalk is the way to express knowledge. I 
fully adhere to that. So, as I'm into Image Processing and Graphics, I 
don't want to call Cairo. I want access to all the code (i.e. Morphic 3).

On the other hand, I'm not into File Systems. I'm not interested into 
having a FAT-32 or whatever file system code in Smalltalk. I'm perfectly 
happy in letting the platform below handle it, if it can. To me a VCS 
can be seen as a versioning file system. That's exactly what Git does, 
and it does it quite well! So, I'm happy to access it, and not duplicate 
its logic in Smalltalk.

Cheers,
Juan Vuletich

>> Does Monticello count as file based? The integrated support in Pharo
>> w/ Nautilus is pretty sophisticated; you can see diffs and things
>> right in the browser. It's pretty cool. It does store the changes in
>> the filesystem, but that's not where most of the version control
>> activity takes place.
>>
> Oops, I should clarify.  By file-based, I meant version control 
> systems that operate directly on files.  I agree Monticello falls 
> outside of this definition, and is cool.