Squeak-dev June 2003

squeak-dev@lists.squeakfoundation.org

188 participants
973 discussions

Re: About 3.6 alpha process: to break the less
by Daniel Vainsencher 03 Jun '03

03 Jun '03

Stephane Ducasse <ducasse(a)iam.unibe.ch> wrote: > Hi > > this would be a pity because we would introduce a lot of errors in > external packages. ... > Now I was thinking that we are doing too much in the same time. > Cleaning in a closed world assumption is much easier > than in presence of removal. But of course you can't work under a closed world assumption - unless you load everything in SM (and probably different things in it clash), you're by definition breaking stuff whenever you change APIs. > So now that we are following this path, I think that the best way to > continue would be to > have a script loading all the packages that will be present in the 3.6 > full image and that the cleaners > works on this image. IMO, the solution is not for you to work with a different subset of stuff than the alpha stream, the solution is along the lines of tests and tools for external package maintainers. And requires SM 1.1, for package version support. > What will be absolutely important for the future is to have package (a > class knowing which class and methods it contains), this way (except > for class extensions made from one package to another one), we could > make that each > change related to a given package goes into the right package. This is orthogonal, but yes, I agree we'll need something like that. PackageInfo can do this, but depends on people adhering to its standards. Daniel

1 0

[VM][BUG] losing alpha channel
by Lex Spoon 03 Jun '03

03 Jun '03

asFormOfDepth: doesn't fixup the alpha chanel on conversion from 16 bits to 32 bits. Similarly, drawing a 16-bit form onto a Form using a FormCanvas has the same problem. Conversion from 8 bits to 32 bits has no trouble. What's up? I notice a method called fixAlpha:with: with a very suggestive comment, but I don't know how to use it. I post this in the hopes someone can suggest a one-line fix, before I delve deeper. My image is 3.5-5174 , though non-standard changes may well have crept in My VM is Ian's 3.4 release Lex ===== form0 := ImageReadWriter formFromStream: 'http://www.squeakland.org/images/logo.jpg' asUrl retrieveContents contentStream. form32 := form0 asFormOfDepth: 32. form32 colorAt: 50@50 "transparent!!" form32 := (Form extent: form0 extent depth: 32). (FormCanvas on: form32) drawImage: form0 at: 1@1. form32 colorAt: 50@50 "also transparent" form32 := (form0 asFormOfDepth: 8) asFormOfDepth: 32. form32 colorAt: 50@50 "this is fine -- the problem seems peculiar to 16->32 conversions"

2 1

fonts and unicode
by Greg Hamilton 03 Jun '03

03 Jun '03

I've been tinkering with Squeak in my spare time and having a whale of a time learning Smalltalk and watching the little mouse and such. I guess I'm probably not the first person to say that the default fonts are horrible. Is it possible to add new, scalable fonts ? Has anybody done any work on antialised fonts ? And what's the state of unicode support in Squeak. I think everything I've looked at so far is stored in ASCII. Greg

5 7

Re: request for critique
by Richard A. O'Keefe 03 Jun '03

03 Jun '03

Concerning efficiency, let's look at one of the methods: iterationsSurvivedBy: aPoint "Answer how many iterations of z(0) = 0 z(i+1) = z(i) squared + aPoint it took to discover that aPoint is not in the Mandelbrot set. Answer -1 if aPoint was not ruled out before the maximum number of iterations was executed." | a b | a := aPoint x. b := aPoint y. 1 to: maxNumberOfIterations do: [:iteration| |aSquared bSquared oldA| aSquared := a squared. bSquared := b squared. aSquared + bSquared > 4 ifTrue: [^iteration]. oldA := a. a := aSquared - bSquared + aPoint x. b := 2 * oldA * b + aPoint y]. ^-1. ! ! I am a little puzzled that the comment says that iterations start with (complex) 0 but the code starts with (aPoint x, aPoint y). I note that having extracted aPoint x and aPoint y, they are extracted again and again and again. This would be more efficient as iterationsSurvivedBy: aPoint |a b x y| a := x := aPoint x. b := y := aPoint y. 1 to: maxNumberOfIteractions do: [:iteration| |a2 b2| a2 := a * a. b2 := b * b. a2 + b2 > 4 ifTrue: [^iteration]. b := 2 * a * b + y. a := a2 - b2 + x]. ^-1 Yes, x*x is marginally faster than x squared (saves a message send). For a calculation like this, eliminating the repeated calls to Point {x,y} and #squared may be worth it, generally one wouldn't bother.

1 0

Re: Celeste and FilteringCeleste
by Daniel Vainsencher 03 Jun '03

03 Jun '03

First I wanted to point out two things you seem to be ignoring - 1. Celeste is no longer in the image. I posted a script for it's removal a while ago, and it was subsequently patched and rolled into the update stream for many of the reasons you mention. 2. Adam Spitz has taken over maintainance of Celeste, and has already integrated a few patches that had been laying around. I would definitely love to see you and Adam work out how to come up with one core Celeste that has the best features of both, maintained by one person. Anyway, now that Celeste is out of the image, it should not be too hard to go ahead and separate the UI from the model (which would not be a bad idea anyway), and then maintain two UI based on the same model, if we find the preference differences are hard to reconcile. Daniel Lex Spoon <lex(a)cc.gatech.edu> wrote: > > I do prefer the UI. The UI is identical except for how it deals with > filtering. Instead of having a button for each type of filter, FC gives > you a generalized list of filters. FC's filters may include the usual > participant and subject filters, plus a new "category" filter to replace > the top-left pane of classic Celeste. The custom filters are still > there, but they've been renamed to "named filters". > > That's it. This UI is still fast to use for the usual case of one > category filter, one named filter, and one subject or participant > filter. Now, however, you can have multiple filters of the same kind if > you want. > > I actually envisioned FC letting you combine multiple filters with AND, > OR, NOT. In particular, it's lame to resort to a "code" filter just > because you want to do a participant filter on multiple possible > participants, e.g. to filter out all email that is addressed to any of > your personal email addresses. Sadly I don't have time to work on stuff > like this. > > Aside from this there are a lot of small goodies that I don't think ever > got into the main image's Celeste. From the top of my head: > > - It is speedy on large folders > > - It interacts correctly with image snapshots > > - It lets you open multiple windows at once on the same mail database > > - It does some save's more quickly, due to putting Daniel's (I think?) > logging idea straight into the index file > > > Then again, I'm sure a patch or two has gone into main Celeste but not > into FilteringCeleste. > > > I never intended for there to be a fork. The thing is that no > individual is able to make major changes to the "main" Celeste. There's > no designated maintainer, Squeak Central doesn't maintain it, and the > harvesters are too conservative. So FilteringCeleste has become my own > release of Celeste, much like Snail on Hot Asphault (SHA) was my own > release of the Unix VM back when Ian Piumarta was busy doing other > things. I happen to think I've done a great job on it :), but it's up > to the community what to do next. I'll personally keep using FC until > there's a better alternative. > > IMHO it is a waste of time to do non-trivial work on "main" Celeste. FC > is superior in many ways, and even the UI can be switched if it is a > real roadblock for people. Celeste only gets modified right now by > going through the harvesters, and this is not a process that allows > major changes. If some maintainer does step up and start working on the > main Celeste, I still think the best thing would be to start by > integrating the improvements FC makes. However, that effort will > probably be more than the effort of adding a classic UI to FC. FC is > simply much more about non-UI things than UI things, despite its name. > > > I never thought I needed to say it explicitly, but I'd certainly be very > happy for FC to be considered "the" maintained version of Celeste. I > would integrate patches -- or negotiate with people until I would accept > the patches -- and I'd make an occasional release. Celeste hackers > would love this, I think; even with my limited time, it should be much > faster to get things posted through me than through the harvesting > process. Further, I can allow much bigger changes to go through than > will make it through into the main image. > > Let me put this another way. Having a single maintainer is a good for > an open source project. A maintainer can move things along more > quickly, and a maintainer can keep the program coherent. I don't insist > on the maintainer being me, but I'm a natural choice and a good choice. > Natural, because I got Celeste into Squeak to begin with and because > I've been maintaining it ever since then. Good, because I know the code > extremely well, I have a good design sense, and I think I'd be pleasant > to work with. While my pickiness (to be kind) about design would irk > some people, I'm quite willing to go through iterations with people. > > On the other hand, I have limited time. If someone more eager wants to > jump in and take over FilteringCeleste, that would be great. I'd like, > however, to at least see some patches from them before handing over the > baton, and I haven't seen a single patch as yet. > > > In sum, both general users and more devoted hackers should be much > happier with Filtering Celeste than the Celeste in the main image. > Switch over, already. :) Squeak Map makes it very easy. > > > Lex

2 1

[BUG][FIX]FileContentsBrowser>browseSenders
by Tim Rowledge 02 Jun '03

02 Jun '03

As of a 5240 image, FileContentsBrowsers need this small change to keep #senders working. tim -- Tim Rowledge, tim(a)sumeru.stanford.edu, http://sumeru.stanford.edu/tim Gotta run, the cat's caught in the printer.

1 0

Doh! (was: Image hanging on startup)
by Larry White 02 Jun '03

02 Jun '03

I wanted to follow up since a number of people offered suggestions and some mentioned having similar image hanging problems: The root cause was that McAfee's NetShield virus protection software conflicted with Squeak while scanning certain versions of my image file as they were loaded into memory. The net result was that they deadlocked and Squeak appeared to hang. Only certain images were affected but any image that hung was usable in spite of retry's, etc. Once Netshield was modified to ignore my squeak directory, versions that I was unable to load became usable once again. I have no idea why some images were impacted and others were not. No issue with Squeak at all. Thanks again for the help received. -larry _________________________________________________________________ Tired of spam? Get advanced junk mail protection with MSN 8. http://join.msn.com/?page=features/junkmail

1 0

Squeakers at ESUG
by Daniel Vainsencher 02 Jun '03

02 Jun '03

Who is planning to go to ESUG? I am, and I was thinking it would be nice to organize our Squeak presence there. At least to have some CDs and maybe shirts there, and depending on how many we are, maybe also organize some demos, bug(fix)-fest, and other nice stuff like that. Daniel

5 5

Check out this dice.com position: 8643DI
by brans＠nerdonawire.com 02 Jun '03

02 Jun '03

I found this great opportunity at Dice, and thought it might be helpful in your job search. http://seeker.dice.com/jobsearch/servlet/JobSearch?op=1002&dockey=xml/e/f/e… ======== For all of us unemployed...

1 0

request for critique
by David Faden 02 Jun '03

02 Jun '03

Hi, I'm interested in getting some comments on a bit of Smalltalk code. Am I missing or misapplying some of the common idioms? Comments on style (or correctness) would be very welcome too. My code is attached to the bottom of this message. The code is for a little Mandelbrot set explorer. It seems pretty slow (compared to similar implementations of the M. set in different environments and compared to other graphics objects within the Squeak image). Is the usual way to create an image pixel-by-pixel in Squeak to create a Form then set the pixels one-by-one using #colorAt: put: or #pixelValueAt: put:? What could I do to improve the performance? Is there a way to associate an hour glass cursor with a single Morph to allow it indicate that it's carrying out a noticeably time consuming process? Is there a simple way to have a Morph just take over the handling of some MouseEvents but not all so that it will get the default handling for the events it's not interested in? Thank you. David

4 4

Jump to page: