Squeak has had a lot on its plate. And as Marcus and Stef deliver 3.9 with a lot of good features, it would be good to let Squeak take time for digestion.
The goal of a refactoring pass is that no behaviour is changed and definitely no new features added. The code is rearranged to make more sense organizationally. And methods that now carry the kitchen sink along with them get to put the sink back in the kitchen where it belongs.
As I try to learn squeak by bug hunting, I am recongizing a lot of bugs are there because of confusion on the part of the would be maintainer. More over I am seeing a lot of code that is hard to track and easy to get confused by. In otherwords hard to maintain.
A breather would be in order.
The smell detecters should be brought out of moth balls. The cobwebs cleaned from the attic, the rugs taken out an beaten. The windows cleaned and opened so everything can be aired out.
It could be just a short iteration. The code needs it. It would work the same at the end as it did at the beginning and we would have a cleaner base from which to add our new improvements.
Yours in service,
Jerome Peace
Addenda: If you will look at Mantis #0002568: How much should warpblt inherit from BitBlt??? You will see where this is coming from and why it is necessary.
Also: What would be really nice IMHO would be to follow this with a short pass dedicated to just the bug fixes that will have accumulated as we stuck to the discipline of a refactoring pass.
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
On 1/24/06, Peace Jerome peace_the_dreamer@yahoo.com wrote:
The smell detecters should be brought out of moth balls. The cobwebs cleaned from the attic, the rugs taken out an beaten. The windows cleaned and opened so everything can be aired out.
Yup. Personally, I think: - Start refactoring Collections and Morphic using Traits - I think these two packages have a lot of code duplication all over the place that could be neatly cleaned up in that way; - Junk Etoys (SqueakLand has made it abundandly clear they're not interested in us maintaing Etoys; we need to see whether it makes sense to maintain a separate Etoys just for Smalland(?)); - Bring all unit tests to green. All of them (so junk that SUnit unit test that is supposed to give red - green is green, not 'all green but one red and you have to check every time whether it is that Sunit test'). I can start running Squeak's on Fire to support that process. - Refactor mercilessly and take no prisoners.
In fact, all of that could be done during 3.9b. After all, refactoring is not changing behavior, by definition :-).
Yup. Personally, I think:
- Start refactoring Collections and Morphic using Traits - I think
these two packages have a lot of code duplication all over the place that could be neatly cleaned up in that way;
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
- Junk Etoys (SqueakLand has made it abundandly clear they're not
interested in us maintaing Etoys; we need to see whether it makes sense to maintain a separate Etoys just for Smalland(?));
Indeed
- Bring all unit tests to green. All of them (so junk that SUnit unit
test that is supposed to give red - green is green, not 'all green but one red and you have to check every time whether it is that Sunit test'). I can start running Squeak's on Fire to support that process.
I would like that!
- Refactor mercilessly and take no prisoners.
In fact, all of that could be done during 3.9b. After all, refactoring is not changing behavior, by definition :-).
We should not. We are exhausted and we should soon step back and only look at fixes.
Stef
On 1/24/06, stéphane ducasse ducasse@iam.unibe.ch wrote:
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
Well, I think that if it's there, we should use it. Two of the larger class libraries can benefit tremendously from Traits. What would be a reason *not* to use it?
We should not. We are exhausted and we should soon step back and only look at fixes.
Ok. That's the team's call, of course :-)
Personally, I'd love to see traits integrated and used to refactor core things like the collections library. There've been a lot of negative comments made about the idea, and few positive ones - I've added a vote on http://de-1.tric.nl/seaside/sqp/list for it. Please, everybody who would like to see this happen, vote!
(Aside: why do I have no vote budget? I suppose it's because I only just created my sqp account?)
Tony
Cees De Groot wrote:
On 1/24/06, stéphane ducasse ducasse@iam.unibe.ch wrote:
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
Well, I think that if it's there, we should use it. Two of the larger class libraries can benefit tremendously from Traits. What would be a reason *not* to use it?
We should not. We are exhausted and we should soon step back and only look at fixes.
Ok. That's the team's call, of course :-)
On 1/24/06, Tony Garnock-Jones tonyg@lshift.net wrote:
(Aside: why do I have no vote budget? I suppose it's because I only just created my sqp account?)
Yup. You need to convince people that you're a nice, hardworking guy so they certify you as something above "Observer", which - as the term says - is seen as someone who is not really in the community, more observing it, so you don't get a lot of rights as Observer on SqueakPeople...
As soon as you get raised to Apprentice (any moment now), you'll have votes.
If you click on the 'Create New' you get the form to create a new issue. A new issue is created even if you do not click the 'submit' button on the new issue form.
Also it appears that my vote budget is zero even though I'm currently certified at Journeyer level.
regards
------------------------ Frank Caggiano frankcag at crystal-objects dot com http://www.crystal-objects.com
The best education for the best is the best education for all. Robert Maynard Hutchins
Thanks for the feedback, Frank - I'll look into it tonight.
On 1/24/06, Frank Caggiano frankcag@crystal-objects.com wrote:
If you click on the 'Create New' you get the form to create a new issue. A new issue is created even if you do not click the 'submit' button on the new issue form.
That neatly explains all the extra entries I've been wondering about :-)
Also it appears that my vote budget is zero even though I'm currently certified at Journeyer level.
You should have 10 votes as Journeyer. You haven't voted already? Is it ok if I login under your account to check?
On Jan 24, 2006, at 10:03, Cees De Groot wrote:
Thanks for the feedback, Frank - I'll look into it tonight.
On 1/24/06, Frank Caggiano frankcag@crystal-objects.com wrote:
If you click on the 'Create New' you get the form to create a new issue. A new issue is created even if you do not click the 'submit' button on the new issue form.
That neatly explains all the extra entries I've been wondering about :-)
Also it appears that my vote budget is zero even though I'm currently certified at Journeyer level.
You should have 10 votes as Journeyer. You haven't voted already? Is it ok if I login under your account to check?
No, haven't voted yet. Sure go ahead and do what you need to do to check it out.
------------------------ Frank Caggiano frankcag at crystal-objects dot com http://www.crystal-objects.com
The best education for the best is the best education for all. Robert Maynard Hutchins
On 1/24/06, Frank Caggiano frankcag@crystal-objects.com wrote:
Also it appears that my vote budget is zero even though I'm currently certified at Journeyer level.
That happens if you login as 'frankcag', which is an alias to 'FrankCag', the real account. Apparently my data model reader isn't completely bugfree, yet.
On 1/24/06, Frank Caggiano frankcag@crystal-objects.com wrote:
If you click on the 'Create New' you get the form to create a new issue. A new issue is created even if you do not click the 'submit' button on the new issue form.
And I think I patched that one as well. With code that cries for refactoring, but i'll leave that for when I have more time...
If you click on the 'Create New' you get the form to create a new issue. A new issue is created even if you do not click the 'submit' button on the new issue form.
That neatly explains all the extra entries I've been wondering about :-)
Seems more like a timeout issue to me because I did submit the one that became blank and it had text.
I, too, don't have my Apprentice level votes.
On 24 janv. 06, at 14:04, Cees De Groot wrote:
On 1/24/06, stéphane ducasse ducasse@iam.unibe.ch wrote:
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
Well, I think that if it's there, we should use it. Two of the larger class libraries can benefit tremendously from Traits. What would be a reason *not* to use it?
- (1) Not a technical one, but I do not want to give the impression that we want to go fast and crazy.
- (2) finish 3.9 :)
- (3) and release our pressure..... :)
- (4) rest a bit
We should not. We are exhausted and we should soon step back and only look at fixes.
Ok. That's the team's call, of course :-)
:)
stéphane ducasse wrote:
Yup. Personally, I think:
- Start refactoring Collections and Morphic using Traits - I think
these two packages have a lot of code duplication all over the place that could be neatly cleaned up in that way;
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
But could we not revert back just by flattening the Traits version? It would seem to be a good excercise anyway. Morphic is (said) to be due for a rewrite, and Traits would appear to be a good method, and the effort would be a good trial of Traits.
David
On 24 janv. 06, at 20:48, David P Harris wrote:
stéphane ducasse wrote:
Yup. Personally, I think:
- Start refactoring Collections and Morphic using Traits - I think
these two packages have a lot of code duplication all over the place that could be neatly cleaned up in that way;
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
But could we not revert back just by flattening the Traits version?
yes we can :)
It would seem to be a good excercise anyway. Morphic is (said) to be due for a rewrite, and Traits would appear to be a good method, and the effort would be a good trial of Traits.
Yes but as I said. We are responsible (or feel it) for 3.9 and we start to be exhausted. Marcus should really focus on his PhD and I should slow down with harvesting. At the end of the day I would like to code something fun :). Or clean just for fun and let other deal with the mess of integrating, complains..... you see just the fame, glory and fun. :)
Stef
stéphane ducasse wrote:
On 24 janv. 06, at 20:48, David P Harris wrote:
stéphane ducasse wrote:
Yup. Personally, I think:
- Start refactoring Collections and Morphic using Traits - I think
these two packages have a lot of code duplication all over the place that could be neatly cleaned up in that way;
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
But could we not revert back just by flattening the Traits version?
yes we can :)
It would seem to be a good excercise anyway. Morphic is (said) to be due for a rewrite, and Traits would appear to be a good method, and the effort would be a good trial of Traits.
Yes but as I said. We are responsible (or feel it) for 3.9 and we start to be exhausted. Marcus should really focus on his PhD and I should slow down with harvesting. At the end of the day I would like to code something fun :). Or clean just for fun and let other deal with the mess of integrating, complains..... you see just the fame, glory and fun. :)
Stef
No, I fully appreciate the amount of work that goes into even (supposedly) simple changes. And I agree that a 'new' project should wait for a recharge of energy.
While I can appreciate the big strokes of these discussions, my knowledge of the insides of Squeak, and Morphic in particular, are extremely weak. I would like to help a rewriting of Morphic, but I am afraid I would only slow it down with my learning curve. Is there an outline of how Morphic should be restructured -- that kind of speculation is fun (maybe).
Again, thanks to you and Marcus. I always found having to concentrate on my PhD was a good way to stimulate other activities :-)
David
st,Aiphane ducasse writes:
Yup. Personally, I think:
- Start refactoring Collections and Morphic using Traits - I think
these two packages have a lot of code duplication all over the place that could be neatly cleaned up in that way;
I would not so that people do not feel prisoner of traits. Our goal is not to impose traits.
I agree that we should be cautious about introducing traits into the base image however we should also have a few good examples of traits use in the image for the rest of us to learn from.
The image has always been an important source of documentation and examples in the Smalltalk world. We do need some widely shared traits examples.
Bryce
On 24 janv. 06, at 01:56, Peace Jerome wrote:
Squeak has had a lot on its plate. And as Marcus and Stef deliver 3.9 with a lot of good features, it would be good to let Squeak take time for digestion.
Indeed we plan to arrive in beta soon after the annotations get in. Currently there are some discussions to have them compatible with VW without breaking the one of Tweak. This would make the life of Seasider easier and as the one of VW are good copying and be compatible would be nice to have.
The goal of a refactoring pass is that no behaviour is changed and definitely no new features added. The code is rearranged to make more sense organizationally. And methods that now carry the kitchen sink along with them get to put the sink back in the kitchen where it belongs.
As I try to learn squeak by bug hunting, I am recongizing a lot of bugs are there because of confusion on the part of the would be maintainer. More over I am seeing a lot of code that is hard to track and easy to get confused by. In otherwords hard to maintain.
A breather would be in order.
The smell detecters should be brought out of moth balls. The cobwebs cleaned from the attic, the rugs taken out an beaten. The windows cleaned and opened so everything can be aired out.
It could be just a short iteration. The code needs it. It would work the same at the end as it did at the beginning and we would have a cleaner base from which to add our new improvements.
Yours in service,
Jerome Peace
Addenda: If you will look at Mantis #0002568: How much should warpblt inherit from BitBlt??? You will see where this is coming from and why it is necessary.
Also: What would be really nice IMHO would be to follow this with a short pass dedicated to just the bug fixes that will have accumulated as we stuck to the discipline of a refactoring pass.
Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
squeak-dev@lists.squeakfoundation.org
-
Bryce Kampjes -
Cees De Groot -
Darius Clarke -
David P Harris -
Frank Caggiano -
Peace Jerome -
stéphane ducasse -
Tony Garnock-Jones