Code-Forking

Satellite Boy intern_boy_chris at yahoo.com
Tue Jul 13 15:26:51 UTC 1999


> 
> From what I know of Linux development (which is not much), a
> 'code fork'
> like such a parallel version is to be avoided at all costs.
> However, there
> is no way that Linus can check and veto all the changes proposed
> for the
> Linux kernel. By analogy, Squeak may get to the stage where Dan
> cannot
> personally examine every proposed change set and its implications
> (I could
> be wrong about this - nothing seems to be beyond Dan and Co). The
> solution
> to this problem which has emerged for Linux development is a kind
> of
> hierarchical delegation. For example, one group will work on
> infra-red
> drivers, and their leader will make recommendations to Linus,
> another will
> work on graphics modules or SMP. Because Linus knows and trusts
> the
> judgement of these group leaders, he no longer has to
> painstakingly check
> every line of code in the proposed additions.
> 
> One problem with Squeak, IMO, is that there are too many people
> (myself
> included) all working independently. This may have something to
> do with the
> history of Smalltalk (it can be difficult to integrate diverse
> images, etc)
> or with the fact that we are still awaiting 'critical mass'.
> (This is why I
> strongly believe we need as many newbies as possible). What I am
> hoping is
> that the better projects (whatever better means) will gain more
> followers,
> and become group projects. The pluggable web server may be the
> first example
> of such a group project, but more of this needs to happen. The
> more groups
> start coalescing around projects, the better. This will also help
> Dan and
> Co., since the groups will become more independent, adding 'final
> products'
> to Squeak that need less intervention by Disney.
> 
> As Linus has said on many occassions, the best thing to do is not
> something
> that appears to be necessary for Linux (or Squeak), but something
> you
> personally will find fun and useful. This way, you will always be
> passionate
> about what you are coding (this is probably the single thing that
> makes open
> source software of such high quality). However, if you need tools
> along the
> way (and who doesn't) then you may find yourself filing in the
> code of
> others. Here's where we can all help each other - if the code has
> shortcomings either let the author know, or contribute an
> improvement to the
> author. If the code is solid and useful, email Dan and ask for it
> to be
> filed in to the base release, or as a standard add-on package.
> Once Dan
> receives 5000 emails requesting something be included, he will
> probably take
> a look at it :-)
> 
> If you are working on something alone, email this group about
> your design
> decisions, problems, workarounds etc. I have received excellent
> code
> suggestions  whenever I have done this (for example a Recursive
> Semaphore
> implementation, and much other stuff). This will also help in
> 'seeding' new
> groups interested in the same code. I guess what I am saying is,
> rather than
> develop a parallel fork of Squeak, lets form groups around the
> code we want
> included in Squeak, test it, improve it, pre-digest it, and send
> it to Dan
> for consideration....
> 
> Anyway, this is turning into a bit of a rant so </rant>
> 
> Peter
> 
> Hi, I'm Chris and I'm a newbie to this list, but I thought since
I know a little bit about linux, I'd offer some suggestions:
	There are development forks. You just don't hear about them. Some
are just people playing around, some are just trying to
Pentium-Optimize linux. Right now, they are not significant.
	Linux is not the only free unix. *BSD was first. This is where a
lot of code-forking occurs. (We have Net-, and Open-BSD. They can
also charge money for BSD software, because its not GPL. MacOS X
Server is based off BSD.)
	However, I believe a code-fork, is not what we need here.
That would risk [one of] Squeak's advantages: cross-platform
compatibility. Or even Cross-Squeak compatibility if the fork was
over some serious. (Like changes in the Object class.)
	The projects thing is a good idea. Personally, I'd like to see FTP
and gopher add to Squeak's list of web tools.
	Heck, if you guy feel this is a problem, email Dan (or whoever the
head guy is.) and get him to set up a "Squeak Development Swiki"
	Those are my thoughts. Even though I'm new to Squeaking, I enjoy
it very much.

_________________________________________________________
Do You Yahoo!?
Get your free @yahoo.com address at http://mail.yahoo.com





More information about the Squeak-dev mailing list