(This seemed sufficiently different from the original thread subject, "FileList[2]>directory: oddity", that I thought I'd start a new thread. Squat mailing list info is at http://netjam.org/squat . Squat is a minimal-snapshot-and-virtual-machine project.)
Hi Stef--
do you have a plan to get a smooth integration of Flow (meaning slowly deprecate the current stream)? If I remember old discussions people were afraid of big bang changes. However it would be good to start to thing how flow could become a good replacement of Stream.
What are your current plans? I'm not expert in Streams so does anybody expertise on Flow and stream comment?
As I've mentioned before on the Squeak list (but not on a webpage, sorry, will get to that), I'm focusing my attention on Flow releases for Squeak 4 and beyond, when ostensibly there will be a minimal snapshot to work from. It's just so much nicer than dealing with the traditional accreted snapshot and the release process constaints it imposes. And I don't just feel that way in the context of Flow; I want to release *everything* I do as modules for a minimal snapshot.
After that infrastructure is in place and Flow is available through it, then I might make releases for older systems. It's not all that hard; I had to do it for 2.2 to get the Squat work started, after all. :) It just takes time, and the minimal snapshot work is much more important to me right now.
So that's timing (although we don't have dates for things yet). As for the technical execution, I think it'd be straightforward. I've been careful in each Flow release so far to make sure that the old and new stuff work together in the same system, although some *clients* may need to be updated, when there is class name contention. The most recent release of Flow, 2 alpha, does this for Squeak 3.2, the version that was current at the time (August 2002). Class name contention was the main issue with that release (the old Stream became "OldStream", etc.). Not really a big deal, in my opinion, and of course handled a lot more easily in a modularized system (where one can refer to classes by ID and not name, for example, and have multiple versions also). I've found in the past that if you let the old things keep the old names, people stay complacent, even if you run reports that point out who hasn't converted yet. :) Finally, the Flow 2a release has everything separated out pretty well into different changesets; in particular, all the forward-compatibility stuff is easy to find.
Another thing that makes me confident is that I've already converted a Squeak system from old streams to new (at a past employer). I actually removed all the old stuff, and things were fine.
thanks,
-C
-- Craig Latta http://netjam.org/resume craig@netjam.org [|] Proceed for Truth!
Hi Stef--
Thanks craig for this email. My gut feeling impression is that 4.0 is far in the future for me.
Does that mean that if there were a minimal-snapshot-plus modules system available right now, it would still be a while before you could use it? Just curious.
So may be if someone is willing to do that it would be a good idea to have Flow sooner.
I actually think it'd be better to obviate any reasons to stay with the accreted snapshot. :) Making major changes in the old snapshot will just slow that down.
I understand that you want to focus on something else, still I'm a bit afraid that this will not happen (but may be this is my dark face that is talking).
Well, it's already happening from a technical perspective. :) I guess there's some question as to whether it'll be Squeak or something else, sure. That's part of the planning I'd like to do now.
thanks again,
-C
-- Craig Latta http://netjam.org/resume craig@netjam.org [|] Proceed for Truth!
On Dec 2, 2003, at 1:21 PM, Craig Latta wrote:
Hi Stef--
Thanks craig for this email. My gut feeling impression is that 4.0 is far in the future for me.
Does that mean that if there were a minimal-snapshot-plus modules system available right now, it would still be a while before you could use it? Just curious.
Depends which modules you have available for it. I don't think many people are going to use a "minimal snapshot" - it's minimal, after all. But can you load Morphic on top of it? What about the Refactoring Browser? How many of the packages on SqueakMap will load, or how much work is it for any one of them to make it into a module?
From your reports, you've done an amazing job stripping down the image, but surely it's an even bigger task to be able to build it back up again?
Hi Avi--
Does that mean that if there were a minimal-snapshot-plus-modules system available right now, it would still be a while before you could use it? Just curious.
Depends which modules you have available for it.
And what you want to do with it. :)
I don't think many people are going to use a "minimal snapshot" - it's minimal, after all.
Sure, but it's attractiveness as a basis for development and deployment is rather a different issue, I think.
But can you load Morphic on top of it? What about the Refactoring Browser? How many of the packages on SqueakMap will load...
I would like each of the authors involved to be interested enough in the architecture of the system to consider making that happen a worthwhile endeavor. That's my motivation for inviting planning discussion now.
...or how much work is it for any one of them to make it into a module?
How much work should it be? Now we're having the discussion. :)
From your reports, you've done an amazing job stripping down the image, but surely it's an even bigger task to be able to build it back up again?
Yes, but it's also much easier to distribute that task over many people, so I think on balance it's a win. I certainly think it's worth doing.
thanks,
-C
-- Craig Latta http://netjam.org/resume craig@netjam.org [|] Proceed for Truth!
Craig Latta wrote:
Hi Stef--
Thanks craig for this email. My gut feeling impression is that 4.0 is far in the future for me.
Does that mean that if there were a minimal-snapshot-plus modules system available right now, it would still be a while before you could use it? Just curious.
I would like one of these the-sooner-the-better.
Especially if you could take one and just add Comanche, SmallWiki, Seaside, etc. and off and running you are with your website.
The full-image Squeak already exists and the minimal-snapshot-plus system becoming available should not impede its growth.
So may be if someone is willing to do that it would be a good idea to have Flow sooner.
I actually think it'd be better to obviate any reasons to stay with the accreted snapshot. :) Making major changes in the old snapshot will just slow that down.
People requiring what the "old snapshot" or current Squeak does can still use the current Squeak. Nothing impairs/impedes current capabilities.
To me this adds to the Squeak/Smalltalk world not subtracts from it.
Ok, ok bad phrasing or maybe not. You decide. :)
It seems this would be of interest to the embedded Squeak guys. Dan, Jon, Michael, ...?
I understand that you want to focus on something else, still I'm a bit afraid that this will not happen (but may be this is my dark face that is talking).
Well, it's already happening from a technical perspective. :) I guess there's some question as to whether it'll be Squeak or something else, sure. That's part of the planning I'd like to do now.
I think many are interested.
Which is easier/better? Comanche/Seaside/etc built-on/converted-to Flow or ... as they are and install Squeak's stream/network classes into minimal-snapshot-plus-modules-image?
Yes, naive question from someone unknowledgeable about it all.
Thanks.
Jimmie Houchin
Hi Jimmie--
People requiring what the "old snapshot" or current Squeak does can still use the current Squeak. Nothing impairs/impedes current capabilities.
Right, I was talking about development speed. It seems like the transition to a modular system will happen faster if there's more attention paid by package authors to making their stuff work in such a system, and each person's time is limited. "Users" probably won't care... until they start wanting to run things in small spaces, or become developers, etc. :)
Which is easier/better? Comanche/Seaside/etc built-on/converted-to Flow or ... as they are and install Squeak's stream/network classes into minimal-snapshot-plus-modules-image?
Or adapted away from the old stream/network classes *and* recast as modules for the minimal snapshot. :) All three seem viable to me (although I admit I'd find the third option most usable :).
thanks,
-C
-- Craig Latta http://netjam.org/resume craig@netjam.org [|] Proceed for Truth!
Hi Stef--
Thanks craig for this email. My gut feeling impression is that 4.0 is far in the future for me.
Does that mean that if there were a minimal-snapshot-plus modules system available right now, it would still be a while before you could use it? Just curious.
Now but right now we saw nothing :) and it takes a lot of times and effort to make think moving. Finally tomorrow a rich company can also offer you to work for it in the dark side :) for a lot of $ (I wish you that). and bye bye squat flow....That's why I do not count in year.
So may be if someone is willing to do that it would be a good idea to have Flow sooner.
I actually think it'd be better to obviate any reasons to stay with the accreted snapshot. :) Making major changes in the old snapshot will just slow that down.
I could not understand that block :)
I understand that you want to focus on something else, still I'm a bit afraid that this will not happen (but may be this is my dark face that is talking).
Well, it's already happening from a technical perspective. :) I guess there's some question as to whether it'll be Squeak or something else, sure. That's part of the planning I'd like to do now.
I understand. The only thing I can tell you is that we are interested. I culd ask guys from here to have a look and buiild stuff on it but they should be able to play with it first. but you decide your plans
thanks again,
-C
-- Craig Latta http://netjam.org/resume craig@netjam.org [|] Proceed for Truth!
Hi Stef--
My gut feeling impression is that 4.0 is far in the future for me.
Does that mean that if there were a minimal-snapshot-plus
modules system available right now, it would still be a while before you could use it? Just curious.
No, but [there haven't been any releases yet] and [it takes a lot of time and effort to make things move].
Yeah. :) Well, I hope that the design notes I've put out are of some help.
Finally, tomorrow a rich company can also offer you to work for it in the dark side :) for a lot of $ (I wish you that). and bye bye squat flow...
Well, all I can say is that I think that's very unlikely. :) I want to be able to keep using my own work, which effectively puts me in the same boat as the public. (Note that, for what it's worth, Flow itself has already been released a couple of times.)
That's why I do not count in year.
I couldn't parse that. :) Do you mean, that's why you don't budget time around it?
I actually think it'd be better to obviate any reasons to stay with the accreted snapshot. :) Making major changes in the old snapshot will just slow that down.
I could not understand that block :)
I mean, currently we fuse (or "accrete") new stuff into the old stuff. That's what I think of as the "old way". I really like having that capability, but I think it's something that makes more sense for a single person. I've always found the notion of a whole community using a particular large object memory to be rather strange, not something I would expect as the default. The larger an object memory gets, the more likely it is to have stuff that someone doesn't want.
Continuing to tweak this large object memory will distract from the task of building a modular system and a set of modules to go with it, as it has distracted us for decades. :)
The only thing I can tell you is that we are interested.
Great! :)
I could ask guys from here to have a look and build stuff on it but they should be able to play with it first. but you decide your plans
Sure. As I mentioned previously, I want Squat to have a decent module-loading-and-unloading demo before I release it, and that's proceeding well.
thanks again,
-C
-- Craig Latta http://netjam.org/resume craig@netjam.org [|] Proceed for Truth!
No, but [there haven't been any releases yet] and [it takes a lot of time and effort to make things move].
Yeah. :) Well, I hope that the design notes I've put out are of some help.
there were exicting sure!
Finally, tomorrow a rich company can also offer you to work for it in the dark side :) for a lot of $ (I wish you that). and bye bye squat flow...
Well, all I can say is that I think that's very unlikely. :) I want to be able to keep using my own work, which effectively puts me in the same boat as the public. (Note that, for what it's worth, Flow itself has already been released a couple of times.)
I know this is why I asked because I would like to avoid to lose good assets.
That's why I do not count in year.
I couldn't parse that. :) Do you mean, that's why you don't budget time around it?
I mean that I do not project myself too far in the future. You know MS could be interest in traits and hop :)
I actually think it'd be better to obviate any reasons to stay with the accreted snapshot. :) Making major changes in the old snapshot will just slow that down.
I could not understand that block :)
I mean, currently we fuse (or "accrete") new stuff into the old stuff. That's what I think of as the "old way". I really like having that capability, but I think it's something that makes more sense for a single person. I've always found the notion of a whole community using a particular large object memory to be rather strange, not something I would expect as the default. The larger an object memory gets, the more likely it is to have stuff that someone doesn't want.
Exact this is why I dream about a mini-image and packages...because even if I value multimedia, I'm a language freak.
Continuing to tweak this large object memory will distract from the task of building a modular system and a set of modules to go with it, as it has distracted us for decades. :)
Yes
The only thing I can tell you is that we are interested.
Great! :)
I could ask guys from here to have a look and build stuff on it but they should be able to play with it first. but you decide your plans
Sure. As I mentioned previously, I want Squat to have a decent module-loading-and-unloading demo before I release it, and that's proceeding well.
So we are waiting :)
thanks again,
-C
-- Craig Latta http://netjam.org/resume craig@netjam.org [|] Proceed for Truth!
squeak-dev@lists.squeakfoundation.org