Hi, how do bugs on mantis get assigned to someone? Who is responsible for this?
YOU DO!!! :)
When you report a bug and assign it a category (which I hope you and everyone else will do, otherwise I have to) that assigns the bug to a team. Currently there is not a team for every category, not even close, nonetheless as these teams are formed they will pick up the categories and since you have already assigned them they will have their bugs assigned to them all in one step.
Ken Causey
On Tue, 2006-03-07 at 18:39 +0200, Dan Corneanu wrote:
Hi, how do bugs on mantis get assigned to someone? Who is responsible for this?
On the subject of Mantis and who does what with it:
It's great when someone takes the time to report a problem and still better when they include a decent amount of information about the problem, how it happened, what image, what machine, what VM, update level etc. Reading http://www.chiark.greenend.org.uk/~sgtatham/ bugs.html is a useful way to spend a moment.
What happens next is..... nothing much, usually. This is a tragedy. All that information waiting to be used to improve our world. Sigh.
So what are you going to do about it? Yes *you*, stop squirming and sit up straight, pay attention and resolve to take your part properly in this process. The simple fact is that for complex problems there are only a few people with the experience to have much likelihood of success and most of them are very busy. Anything that can be done to save time will improve the chance of a complex problem getting tackled and *that* can be done by almost any of us. In the process of the simple jobs I am about to suggest we are going to learn and eventually become expert enough to tackle bigger works, maybe.
Job 1: making sure the report makes sense Just about anyone (even me on a monday morning pre-caffeine) can check a random report, see if it has even faintly enough information to mean anything and decide if it has a hope of being useful. If data is lacking, try contacting the original reporter to see if they can offer more grist for the mill. If they can't, or won't, or say they can reproduce it then we can only sensibly close the report. If you can find related seeming reports using the mantis search facility, link them together.
Job 2: making sure it's reproducible Just about anyone (blah blah) can try to reproduce a problem. If there seems to insufficient information to be able to, see Job 1. Assuming you can reproduce the problem you can add your experience to the report. If it took some time to get to the problem, consider saving an image/changes/ script for future reference.
Job 3: assign it to someone plausible A lot of the time it is quickly obvious where the problem lies and you can assign the report to the right people. If you don't know the right assignee, change the report status to feedback and assign it to your best guess. It's not like you're ordering the person to do work. What we could do with in the Mantis setup is a more useful list of assignable names; I'd really prefer to see 'VM guys', 'Collections folks', 'Windows weenie' than 'fred', 'jkr' etc. Roles are better than names for this purpose.
The nice thing is that you don't have to do all three of these. If you are a real newcomer with an urge to help, just do a bunch of Job 1. If you can achieve a report that makes sense to you as a newcomer then it ought to be really clear to someone more experienced. If you can't make time to test stuff but know a good bit about what goes where and who deals with it, do a bunch of Job 3.
Right now there's only 289 reports listed on the Squeak mantis board. If everyone that was eligible to vote in the recent board elections (340-something?) took a look at doing one of the above jobs per day - forget about actually working on the bug in detail - then within a week we could have every report checked on by several sets of eyes.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: LC: Lobotomize CPU
On 07.03.2006, at 19:55, tim Rowledge wrote:
On the subject of Mantis and who does what with it:
It's great when someone takes the time to report a problem and still better when they include a decent amount of information about the problem, how it happened, what image, what machine, what VM, update level etc. Reading http://www.chiark.greenend.org.uk/ ~sgtatham/bugs.html is a useful way to spend a moment.
What happens next is..... nothing much, usually. This is a tragedy. All that information waiting to be used to improve our world. Sigh.
What I would like to try once would be managing those issues on mantis "GTD" style... I suppose everyone on this list knows what GTD is... no? So there are those kinds of books that nobody admits of reading but magically they sell a lot: Self Help Books. (I hear a terrified scream from the back). GTD is just one of those books... but it was somwhow good enough to get it's own TLA. Now that says something about hacker-compatibility...
The first step to GTD is to take the amorphous blob (Mantis) and declare it to be the INBOX. Now let's take every item out of the INBOX and ask one simple Question: "What is the next action?".... and so on. There are lots of descriptions of the process on the web (just google for GTD to learn more).
It would be interesting to see how this process would work out... the main problem with most items in Mantis is that we exactly have not defined what need to happen next. Does it need to be discussed? Routed to a Team? put into 3.9a? or 3.8? or both? patched to be able to be added to 3.9a?
Hard to say with the current setup. And then all the stuff that I can't do anything about, why do I see that? For me, a Graphics bug is done as soon as I have recognized it to be a graphics bugs. i can't do anything. move along. Same with network, VM, lots of stuff. There are thigs where the issue needs to be discussed. Others are Projects, we know that they should be done, but it will take a huge effort: for bugfixing, there is nothing that can be done. Then there is the general difference between "bugs that need to be fixed" and "bugs that are fixed but the fix is not yet in the release". Other issues have a huge discussion, mayeb 5 different changesets, some containing tests, some different versions of a fix. What is the next action?
It would be nice to filter a true "next actions" list out of that big pile of stuff that we have now on mantis.
Maybe we can start simple: Define "Next action" for each and every bug report, just as a comment. That would be a start.
Marcus
Job 3: assign it to someone plausible A lot of the time it is quickly obvious where the problem lies and you can assign the report to the right people. If you don't know the right assignee, change the report status to feedback and assign it to your best guess. It's not like you're ordering the person to do work. What we could do with in the Mantis setup is a more useful list of assignable names; I'd really prefer to see 'VM guys', 'Collections folks', 'Windows weenie' than 'fred', 'jkr' etc. Roles are better than names for this purpose.
It seems I can't change assignment for my bugs. They are automatically affected to Ken. Do I need some permissions ?
On Wed, 2006-03-08 at 22:14 +0100, Damien Cassou wrote:
Job 3: assign it to someone plausible A lot of the time it is quickly obvious where the problem lies and you can assign the report to the right people. If you don't know the right assignee, change the report status to feedback and assign it to your best guess. It's not like you're ordering the person to do work. What we could do with in the Mantis setup is a more useful list of assignable names; I'd really prefer to see 'VM guys', 'Collections folks', 'Windows weenie' than 'fred', 'jkr' etc. Roles are better than names for this purpose.
It seems I can't change assignment for my bugs. They are automatically affected to Ken. Do I need some permissions ?
As I answered for someone else recently: You don't directly assign bugs, you set the Category field to the appropriate value which assigns it to an appropriate team.
Ken
P.S. If you don't choose a category and leave it as Any then it is assigned to me so that I'm notified and then I can go in and set the category.
On 8-Mar-06, at 1:21 PM, Ken Causey wrote:
On Wed, 2006-03-08 at 22:14 +0100, Damien Cassou wrote:
Job 3: assign it to someone plausible A lot of the time it is quickly obvious where the problem lies and you can assign the report to the right people. If you don't know the right assignee, change the report status to feedback and assign it to your best guess. It's not like you're ordering the person to do work. What we could do with in the Mantis setup is a more useful list of assignable names; I'd really prefer to see 'VM guys', 'Collections folks', 'Windows weenie' than 'fred', 'jkr' etc. Roles are better than names for this purpose.
It seems I can't change assignment for my bugs. They are automatically affected to Ken. Do I need some permissions ?
As I answered for someone else recently: You don't directly assign bugs, you set the Category field to the appropriate value which assigns it to an appropriate team.
Ah - I see. That's actually a pretty good match to assigning to a role then, which is nice. So let's revise Job 3 to
Job 3: assign it to someone plausible A lot of the time it is quickly obvious where the problem lies and you can assign the report to the right people. The best way to do this is to choose a plausible seeming entry from the Category list which in effect assigns the issue to a team.
There's also a useful meta-Job that anyone can do; look around for reports that seem to be abandoned and see if you can rattle them around a bit. Email the people who reported or commented on it; ask if there is still any interest. Make a nuisance of yourself - most of us can do *that* well enough.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: TVO: Type Various Obscenities
Here's what I do when I find a bug in squeak:
1. 90% of the time fix it myself without telling anyone, the rest I can't fix and I try to report...
2. I play around with squeak's built-in reporting system and get yelled at to post it on Mantis instead...
3. Scratch my head and wonder what a carnivorous insect has to do with my bug...
4. Wait 4 days...
5. In the middle of a dream, while my mind is making random associations, that mantis is actually squeak's bug tracking system.
6. go to one of the well known squeak sites such as squeak.org looking for a convenient link to the site. -- which fails...
7. google the damn thing yet again... (except that I'm beyond caring at this point...)
Alan Grimes alangrimes@starpower.net wrote:
Here's what I do when I find a bug in squeak:
[SNIP of blablabla]
You forgot:
8. Bitch on squeak-dev instead of: a) helping WebTeam to improve squeak.org b) help fixing the in-image-bug-reporting-mechanism to point to suggest mantis c) ask how to help in any other way or (gasp) actually just do it
/Göran
PS. Not on the board anymore so I can actually write a post like this! Nice. Ok, so I am in a slightly bad mood for the moment - sue me. I will be friendlier when I get more sleep. ;)
On 7-Mar-06, at 4:06 PM, goran@krampe.se wrote:
PS. Not on the board anymore so I can actually write a post like this! Nice. Ok, so I am in a slightly bad mood for the moment - sue me. I will be friendlier when I get more sleep. ;)
Nobody told me I'd have to be *nice* just because I'm on the board. I don't *think* so! One of the few privileges remaining to the aristocracy is being able to be rude and have it count as eccentricity....
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Useful random insult:- Moves his lips to pretend he's reading.
goran@krampe.se wrote:
Alan Grimes alangrimes@starpower.net wrote:
Here's what I do when I find a bug in squeak:
[SNIP of blablabla]
You forgot:
- Bitch on squeak-dev instead of: a) helping WebTeam to improve squeak.org b) help fixing the in-image-bug-reporting-mechanism to point to suggest
mantis c) ask how to help in any other way or (gasp) actually just do it
Okay, I can play this game.
I was about ready to concede when I remembered what happens when I do try to contribute.
A fair number of moons ago, I tweaked the Sokoban Morph to within an inch of it's life and tried to publish it on Squeak's version control archive (I can't remember it's name...). While some might take issue with some of my stylistic decisions, the code did work and demonstrated a clear improvment in the quality of gameplay. (Implementing my changes from scratch would take about 20 hours of work..) I submitted this to whom I thought was the maintainer. He liked some of my changes but preferred a different visual style... I'm really not sure what happened to that version... A few weeks later a version of Sokoban appeared in the mainline. It was based on the same suckey code that had inspired me to fix it in the first place!
That isn't to say that none of my changes are appreciated. I made several enhancements to B3DMatrix (If I remember the class name correctly...) that either simplified the call graph, removed redundant methods, or helped reign in excessive GCing... One person noticed one of my changes and is probably enjoying it on his system right now. The rest of you aren't.
Contribute to Squeak? What's the point... Your enhancements and fixes vanish into the aether and never get used.
Squeak is still a great system but all the cool things are happening behind closed doors at the croquetproject... Ofcourse when 1.0 comes out I'll have to back-port in all of the enhancements I've made to 0.3. =\
Hi alan
you should not mixed Squeak and Sokoban or any goodies. Have you contacted his author robert hirsfeld May be is simply did not like your changes?
Stef
On 8 mars 06, at 03:57, Alan Grimes wrote:
goran@krampe.se wrote:
Alan Grimes alangrimes@starpower.net wrote:
Here's what I do when I find a bug in squeak:
[SNIP of blablabla]
You forgot:
- Bitch on squeak-dev instead of: a) helping WebTeam to improve squeak.org b) help fixing the in-image-bug-reporting-mechanism to point to
suggest mantis c) ask how to help in any other way or (gasp) actually just do it
Okay, I can play this game.
I was about ready to concede when I remembered what happens when I do try to contribute.
A fair number of moons ago, I tweaked the Sokoban Morph to within an inch of it's life and tried to publish it on Squeak's version control archive (I can't remember it's name...). While some might take issue with some of my stylistic decisions, the code did work and demonstrated a clear improvment in the quality of gameplay. (Implementing my changes from scratch would take about 20 hours of work..) I submitted this to whom I thought was the maintainer. He liked some of my changes but preferred a different visual style... I'm really not sure what happened to that version... A few weeks later a version of Sokoban appeared in the mainline. It was based on the same suckey code that had inspired me to fix it in the first place!
That isn't to say that none of my changes are appreciated. I made several enhancements to B3DMatrix (If I remember the class name correctly...) that either simplified the call graph, removed redundant methods, or helped reign in excessive GCing... One person noticed one of my changes and is probably enjoying it on his system right now. The rest of you aren't.
Contribute to Squeak? What's the point... Your enhancements and fixes vanish into the aether and never get used.
Squeak is still a great system but all the cool things are happening behind closed doors at the croquetproject... Ofcourse when 1.0 comes out I'll have to back-port in all of the enhancements I've made to 0.3. =\
-- Don't let your schoolwork get in the way of your learning.
Squeak is still a great system but all the cool things are happening behind closed doors at the croquetproject... Ofcourse when 1.0 comes out I'll have to back-port in all of the enhancements I've made to 0.3. =\
Just curious ... what improvements have you been able to make to Jasmine?
Also, you may be interested to know that I am working on the Croquet Consortium public portal, which will, among services, provide SVN for Croquet's code and communal content after 1.0 is released, and the development model moves out of the cathederal and into the bazaar.
-- Ed
Just curious ... what improvements have you been able to make to Jasmine?
3D models use alot of matrix multiplies... Usually this is done with three physical matrices. Each multiply creates a new matrix for the result... If the code does many many multiplies such as a multi-jointed 3D model and camera motions, then quite a bit of memory gets burned, (8x16 bytes + headers for each discarded matrix). I wrapped one of the primatives and made a "multiply and discard" method which overwrote one of the operands with the result. This new code can replace many instances of the old 3-way call hence reducing the time spent in the GC...
I also identified two methods in the matrix class which did the exact same thing, I removed the one with the larger call graph and re-named the other. (it had fewer senders...)
Most of my changes dealt with these two items...
/ me tried to exparament with Big Mesh teapot but ran into the flaming signed pointer bug... =P
the video drivers I curently have installed has some texture bugs too. =\
reviewing my changes, I see I did a lot of scope demotions too. Unfortunately the compiler doesn't seem to honor scope in such a way that it could be a performance boost. =\ (If a variable is only used in a conditional branch, it doesn't need to exist unless the branch is executed... )
I attached the version I currently have on my system. Note the removed methods.
I'm very sorry, I didn't mean to send 40k to the list. I frequently correspond with Mr. Boyce Off-list and was too tired to notice the return address...
I see I have managed to kick off some discussions on the issue of bugs handling / fixing. This is gooood ...
Alan Grimes wrote:
Here's what I do when I find a bug in squeak:
- 90% of the time fix it myself without telling anyone, the rest I
can't fix and I try to report...
- I play around with squeak's built-in reporting system and get yelled
at to post it on Mantis instead...
- Scratch my head and wonder what a carnivorous insect has to do with
my bug...
Wait 4 days...
In the middle of a dream, while my mind is making random
associations, that mantis is actually squeak's bug tracking system.
- go to one of the well known squeak sites such as squeak.org looking
for a convenient link to the site. -- which fails...
Actually there is a link on squeak.org http://www.squeak.org/Community/ Anyhow, I would say it would be great to have a link to the Bug Database in the Links panel (on the right side). I have found myself looking for it a few times...
- google the damn thing yet again... (except that I'm beyond caring at
this point...)
I know this thread is a bit stale now, but didn't see what the likely conclusion of this issue is meant to be
I see that 3.9 (7055) displays underscores - I installed the Shout package (to get syntax highlighting back) and see that there is a preference for underscore vs :=. Is it the idea that the appearance of the left arrow will eventually be built into the code editor, and for the interim those who prefer left-arrow should use bear with the underscore actually appearing in code?
Would like to vote that that the the elegance, and combined millennia's worth of neural wiring among smalltalkers who've become accustomed to the left-assignment key far outweighs the danger of off-putting newbies.
If it's too difficult to work it into 3.9 for technical reasons, then no complaints there (especially since I'm not volunteering to make it happen) - just don't think that purported "confusion" is a good reason to deny use of left-arrow as an option. I strongly disagree with the idea that having Paragraph editor "magically" display one or the other is somehow a bad thing. Just my $0.02.
Jay
On 6/7/06, Todd Blanchard <tblanchard@...> wrote:
I would also suggest that the left arrow is off-putting to
newbies. There's
no obvious key for it, and it causes a lot of confusion for the
user trying
to figure out how to make one. So it ends up being a barrier to
entry.
Making ParagraphEditor "magically" replace things (or render them
specially)
isn't really better.
Even worse, the tranditional assignment operator looks an aweful lot like the figure that is silk screened on my "Backspace" key (at least on my keyboard). ;-)
I think it's time to let this convention go - it's more trouble
than its
worth.
With that said, I really *love* the single left arrow for assignment. I am saddened by the trend to eject it, but I understand it might not be the best for the long run.
I guess I can get used to := instead, but it just feels so "programmery".
Regards,
John
-- The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore, all progress depends on the unreasonable man. -- George Bernard Shaw
Jay Hardesty schrieb:
I know this thread is a bit stale now, but didn't see what the likely conclusion of this issue is meant to be
I see that 3.9 (7055) displays underscores - I installed the Shout package (to get syntax highlighting back) and see that there is a preference for underscore vs :=. Is it the idea that the appearance of the left arrow will eventually be built into the code editor, and for the interim those who prefer left-arrow should use bear with the underscore actually appearing in code?
Would like to vote that that the the elegance, and combined millennia's worth of neural wiring among smalltalkers who've become accustomed to the left-assignment key far outweighs the danger of off-putting newbies.
If it's too difficult to work it into 3.9 for technical reasons, then no complaints there (especially since I'm not volunteering to make it happen) - just don't think that purported "confusion" is a good reason to deny use of left-arrow as an option. I strongly disagree with the idea that having Paragraph editor "magically" display one or the other is somehow a bad thing. Just my $0.02.
The arrow glyph is still in the (strike-) fonts:
Character value: 16r8F
I guess one could make Shout display that instead of the underscore.
- Bert -
Taking this a step further - anyone know how to configure a keyboard to do this (shift + underscore key) in just one button press?
Better still, any keyboard makers reading this? Email me about my ultimately configurable keyboard idea! Or don't. Suit yourself... ;o)
Jay Hardesty wrote:
Would like to vote that that the the elegance, and combined millennia's worth of neural wiring among smalltalkers who've become accustomed to the left-assignment key far outweighs the danger of off-putting newbies.
Zulq Alam wrote:
Better still, any keyboard makers reading this? Email me about my ultimately configurable keyboard idea! Or don't. Suit yourself... ;o)
Try this one:
http://www.artlebedev.com/everything/optimus/
Each key is an lcd display ;-)
Cheers, - Andreas
Andreas Raab wrote:
Try this one:
http://www.artlebedev.com/everything/optimus/
Each key is an lcd display ;-)
Sweeeeet. I can't tell whether i can have a <- key that sends shift + underscore when pressed - just seems you can map them to programs or single buttons?
(My idea is a little more configurable, though yes, it did have keys like kinda like this albeit less sweeeet)
Hi folks,
I am still waiting for a radically optimistic "wiki like" environment, where anybody (ok, say having at least the status of an apprentice on Squeak-People...) can change code of the current golden_and_shared_image anytime, ideally _always_ developing with it as the base for his (do we have a she on Squeak-People with at least apprentice status?) daily work.
A first rough description for a simple code-wiki architecture I'd like to use: ============= 1.) Some sandbox E.g. Accessing the file system from code that gets pushed into my image is a big no.
2.) Wiki features: Login, History, Changes, Rollback, Links To - Login with squeak-people account - Automatic versioning of methods and classes: When I save a method/ class in my browser, it gets stored as a new version in the central repository. (a tweaked squeaksource?) The new version is only pushed into other clients, when I release the package it resides in. (see 3.)). Other people can see it though if they browsed "versions" of the class/method. - History, changes and rollback of methods, classes and packages (or PackageInfo's as they somehow call them today) - Links to: Who needs that method, class, package
3.) Difference to wikis or Envy - A package knows whether it belongs to the _golden_and_shared_image or not - Packages do not get versioned, they only get released. As an airline company told me recently about my ticket: "Use it - or lose it" ;-) - Methods and classes get versioned automatically each time one changes/saves them, but they do not get released. - Methods and classes know whether they are part of a released package. - If a package gets released - it automatically gets a unique release id - it internally stores all release ids of all required packages (thus serving as some kind of loadable "config map" at the same time) - it gets pushed to every client using the shared environment (only if it belongs into the _golden_shared_image_ of course)
4.) A Notion of quality - Having the idea that any apprentice could push a new release of some package into my current image might naturally lead to more test awareness. One could think about being able to release a package only if -at least all changes (to start from somewhere) are also covered by a green test. -all tests of all _depending_ packages also still work, whether they are in the _golden and shared_ image or not -The more a package is down to the system, the more other packages will depend on it, and the higher the status (apprentice, journeyer or master) of the changing person must be. -Having the notion of package ownership the owner shall definitely at least automatically get a mail, if a new version of his/her package was released
PLEASE: Feel free to play with these ideas (or feel free to let your students play with it) or even implement them- I didn't even have time to write them down... ;-) There are some efforts to do sth like that in Croquet I have heard, but I don't know anything specific.... =================== Btw: Ward Cunningham is working for the "Eclipse committer community" and not for MS since Oct 2005:
"To date, the efforts of the Eclipse Foundation in support of the committer community have been primarily around providing infrastructure and process. However, a high functioning committer community is about more than just sharing servers and following a common process. A high functioning committer community is about collaboration and cooperation between the project silos. Although the Councils do an admirable job of co-ordinating the activities of the many Eclipse projects, what is needed is a culture of collaboration and cooperation. This is especially true today, as Eclipse grows rapidly with new projects and new committers.
To help cultivate this committer culture, Ward Cunningham is joining the Eclipse Foundation as Director, Committer Community Development. Ward's track record of invention in areas such as wikis, patterns and agile development are known worldwide. His current interests in open source and developing communities of developers are a perfect match for the work we need to do at Eclipse. Ward will lead the effort to create a more cohesive Eclipse committer community by working with developers in order to enhance Eclipse as "the place to be".
http://www.eclipse.org/org/press-release/20051019cunningham.html
So let's be faster than them :-)
Cheers,
Markus
On Mar 8, 2006, at 1:37 AM, Alan Grimes wrote:
Here's what I do when I find a bug in squeak:
- 90% of the time fix it myself without telling anyone, the rest I
can't fix and I try to report...
- I play around with squeak's built-in reporting system and get
yelled at to post it on Mantis instead...
- Scratch my head and wonder what a carnivorous insect has to do with
my bug...
Wait 4 days...
In the middle of a dream, while my mind is making random
associations, that mantis is actually squeak's bug tracking system.
- go to one of the well known squeak sites such as squeak.org looking
for a convenient link to the site. -- which fails...
- google the damn thing yet again... (except that I'm beyond
caring at this point...)
-- Don't let your schoolwork get in the way of your learning.
On 8-Mar-06, at 7:05 AM, Markus Gaelli wrote:
Hi folks,
I am still waiting for a radically optimistic "wiki like" environment, where anybody (ok, say having at least the status of an apprentice on Squeak-People...) can change code of the current golden_and_shared_image anytime, ideally _always_ developing with it as the base for his (do we have a she on Squeak-People with at least apprentice status?) daily work.
Given the number of mistakes even the best of the Master rated people make I'd really not like to live in a world where code could just be stuffed in by anyone. Shudder. I mean for goodness sake, I've been doing this continuously for 23 years and still screw up.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: SDP: Search and Destroy Pointer
On Mar 8, 2006, at 8:36 PM, tim Rowledge wrote:
On 8-Mar-06, at 7:05 AM, Markus Gaelli wrote:
Hi folks,
I am still waiting for a radically optimistic "wiki like" environment, where anybody (ok, say having at least the status of an apprentice on Squeak-People...) can change code of the current golden_and_shared_image anytime, ideally _always_ developing with it as the base for his (do we have a she on Squeak-People with at least apprentice status?) daily work.
Given the number of mistakes even the best of the Master rated people make I'd really not like to live in a world where code could just be stuffed in by anyone. Shudder. I mean for goodness sake, I've been doing this continuously for 23 years and still screw up.
23 years of experience and still making mistakes? Tsss. ;-)
The release process for such a "turnaround image"(*) into a stable version could still be conservative: Only allow bugfixes in the last two weeks before a release or so. Hundreds of users will take care. So what would be the worst case scenario besides Homer Simpson using a turnaround image instead of a released one in his nuclear power plant?
Sure, if my image hangs due to some new code somebody on the other end of the world just released, I also get...aehm... sad. Maybe some clever rollback mechanism could be established, otherwise I'd say "no risk no fun" is the mantra of the early adapter.
But I guess the people on the other end of the world will think twice about doing an unsafe new release for a package, as they have to sign it. Trust the fine crowd of squeak developers and they might even reward you by writing tests. People prefer to be proud of their work and not embarrassed, so on the average it should be ok. And if not: Fail early, fail often. Let's dare it and embarrass ourselves from time to time for real - not only here on the mailing list. ;-)
I think as long as nobody has implemented such a "turnaround image" together with a viable migration strategy / hybrid strategy to keep the current dev-process in sync, we don't need to go into further details here. I just wanted to throw some blue pane idea in - maybe it was no the best time for it right now... ;-) People using Mantis and co. are doing a terrific job! And this process seems to be the best we have at the moment. Please, next time also send a link (again) together with your nice motivating words about using mantis. This will hinder people like me from lifting off into blue air and instead make them doing some real pink work. ;-)
Power to the people!
Markus (*) Turnaround: The faster the wheel spins the more stable it gets.
Markus Gaelli gaelli@emergent.de wrote:
it as the base for his (do we have a she on Squeak-People with at least apprentice status?)
Yes: http://people.squeakfoundation.org/person/asparagi/
Regarding the real subject at hand - I have also been thinking in this direction from time to time and I actually have an idea - but will write it up ... this weekend. :)
regards, Göran
squeak-dev@lists.squeakfoundation.org
-
Alan Grimes -
Andreas Raab -
Bert Freudenberg -
Damien Cassou -
Dan Corneanu -
Ed Boyce -
goran＠krampe.se -
Jay Hardesty -
Ken Causey -
Marcus Denker -
Markus Gaelli -
Ron Teitelbaum -
stéphane ducasse -
tim Rowledge -
Zulq Alam