Hello,
what's the current procedures/policies for accepting changes into Squeak?
Back in January of 2001 I submitted an enhancement of the ConnectionQueue class (http://swiki.gsug.org:8080/sqfixes/1285.html). I never received any response.
I'm bringing this up because looking at the list of fixes at:
http://www.mindspring.com/~dway/smalltalk/sqfixes/old/23Aug01/sqfixes4.html
I get the impression that most fixes go unnoticed/uncommented and presumably uncommitted to the official Squeak.
I suppose my enhancement may have contained a bug / been considered badly written; but I would at least have expected to receive some sort of response. Especially since the enhancement addressed the need for blocking accepts - a very basic feature in most development environments. Yet the Squeak Socket/ConnectionQueue stuff only supports a polling approach. In other words, I would have expected the functionality to be desired.
My question is - did my enhancement slip through the cracks, or are the maintainers simply too busy to handle all the changes that come in? Is this a general problem?
If the latter - perhaps a reorganization is needed. There is much talk about Squeak being Open Source and how easy it is to contribute to the project; but if changes go ignored, the submitter is less likely to continue submitting enhancements.
I get the distinct impression that this is a general problem. Surfing the Wiki one can find various nice things (such as better-looking themes, a new and improved system browser, etc) - but they are not in the official Squeak release even though they are seemingly things many people would want - and particularly things that would help Squeak achieve success among newcomers.
Comments? Flames? :)
On Saturday, June 29, 2002, at 09:27 AM, Peter Schuller wrote:
Hello,
what's the current procedures/policies for accepting changes into Squeak?
You submit them here. From time to time they are collected, usually before a new rev is released. Some get accepted for the core image, some don't. Some get listed and saved as "goodies," others don't. Most properly submitted submissions get posted to the Swiki.
Back in January of 2001 I submitted an enhancement of the ConnectionQueue class (http://swiki.gsug.org:8080/sqfixes/1285.html). I never received any response.
The URL reflects that it was "collected" in the Swiki. This is as much response as is likely to happen before a version collection. Usually when fixes are being formally collected, an inquiry goes out for "reminders" that something was overlooked.
I suppose my enhancement may have contained a bug / been considered badly written; but I would at least have expected to receive some sort of response.
Not likely, here or in many OSS projects.
My question is - did my enhancement slip through the cracks, or are the maintainers simply too busy to handle all the changes that come in? Is this a general problem?
Not really, although folks do comment as you have from time to time, both here and in many other OSS projects. Thanks for the contribution, by the way. As a general matter, this "problem" is going to be obviated after general adoption of the new Module systems, in which every user is empowered to post Modules in the Squeak People path.
There is much talk about Squeak being Open Source and how easy it is to contribute to the project; but if changes go ignored, the submitter is less likely to continue submitting enhancements.
Squeak is open source and it is easy to contribute to the project. However, not every submission gets committed to the core image. Many of my submissions were included, others were not.
I get the distinct impression that this is a general problem. Surfing the Wiki one can find various nice things (such as better-looking themes, a new and improved system browser, etc) - but they are not in the official Squeak release even though they are seemingly things many people would want - and particularly things that would help Squeak achieve success among newcomers.
That is why they are on the Swiki -- this is one of the primary places for inclusion other than incorporation into the "core image." There is a strong current prejudice against image bloat, with a focus on refactoring the image for breakout into modules.
[sorry about the late response, been *very* busy]
You submit them here. From time to time they are collected, usually before a new rev is released. Some get accepted for the core image, some don't. Some get listed and saved as "goodies," others don't. Most properly submitted submissions get posted to the Swiki.
Ok. About what my understanding of it was.
The URL reflects that it was "collected" in the Swiki. This is as much response as is likely to happen before a version collection. Usually when fixes are being formally collected, an inquiry goes out for "reminders" that something was overlooked.
Fair enough.
I suppose my enhancement may have contained a bug / been considered badly written; but I would at least have expected to receive some sort of response.
Not likely, here or in many OSS projects.
Usually a person submitting a patch is in some way informed of wheather or not it was comitted by a maintainer.
Not really, although folks do comment as you have from time to time, both here and in many other OSS projects. Thanks for the contribution, by the way. As a general matter, this "problem" is going to be obviated after general adoption of the new Module systems, in which every user is empowered to post Modules in the Squeak People path.
Where can I read more about the module system? It sounds like a good solution. Would this be related to the "virtual root module" window in the latest 3.3alpha?
That is why they are on the Swiki -- this is one of the primary places for inclusion other than incorporation into the "core image." There is a strong current prejudice against image bloat, with a focus on refactoring the image for breakout into modules.
I definitely don't want to bloat the image. I'm much for a modular approach; the monolithic nature of Smalltalk is one of its disadvantages IMO (although there are also advantages to the image approach).
However, there is lot's of not-strictly-necessary stuff in the image. I am not proposing that anything I will mention just now should not be in the image; but if a basic feature such as blocking accepts in the network support is considered bloat, then surely a 3D engine or a pluggable web server is even more bloat?
Again, I am not saying I have anything against the 3D stuff or the web server to be in the image (although a modular approach would be best - for everything), but the bloat argument doesn't really explain why a very basic feature like blocking accepts didn't make it into the image. I mean, we are talking about a feature whose absence in a language like Java would result in the language being considered severely defective!
Reading my changes now, I can see that I am capable of producing a better version of it with my current Smalltalk skills. However, I don't feel particularly inclined to, since I have no indication of why the original patch was not comitted in the first place. And hence, I have no reason to believe an updated version would be accepted. And sine the original ones gets the job done, I don't *personally* need a better version.
Plus, my original patch is already out there. As a result, there is potential for clashes unless I change method names just to avoid it. (I don't particularly believe anyone's actually using my original patch - but how would I know?).
Another thing I don't like about Squeak's networking with respect to this is that accepts are done through polling under the hood anyway, regardless of my patch. This is something I'd like to fix. But it would probably take me a while (not being an expert Squeak:er). However, I am reluctant to try since I have little faith that such a fix would be accepted even if it were to turn out good. And I *don't* want to end up with 75 private patches that never make it into the official Squeak, and that I therefore have to keep comitting to my images after every upgrade - possibly with conflicts every once and a while. (Conflicts are bound to occur because those 75 patches would be modifying "base" classes. Other modifications done will be done relative to the officially release version. If they conflict with my private changes, my changes will break and I would have to fix it.)
Ah well. Looking forward to that modular system possible making this easier!
Where can I read more about the module system? It sounds like a good solution. Would this be related to the "virtual root module" window in the latest 3.3alpha?
I Googled my way there. This is *great*! A giant leap for Squeak IMO - great news, and great job on the part of the authors. Thanks! :)
On Thursday 04 July 2002 04:51 am, Peter Schuller wrote:
Usually a person submitting a patch is in some way informed of wheather or not it was comitted by a maintainer.
I find out the status in one of two ways: * either I see my CS in one of Dan's update announcements * or I look at Bob's Super Swiki http://209.143.91.36/super/415
squeak-dev@lists.squeakfoundation.org
-
Andrew C. Greenberg -
Ned Konz -
Peter Schuller