Squeak-dev April 2000

squeak-dev@lists.squeakfoundation.org

141 participants
506 discussions

RE: VNC Server
by Bert Freudenberg 04 Apr '00

04 Apr '00

On Tue, 4 Apr 2000, Stephen Pair wrote: > Yes, I tried that last night, and it seemed to work very well. Great! There might even be a speedup possibility because in BitBlt you sometimes *know* you're just filling a rectangle, so you don't need to send details over the wire ... > I still need an implementation of the VNC bitmap compression > algorithms (RRE, CoRRE, and Hextile) before the performance will be > even remotely close to acceptable. According to the RFB whitepaper hextile is the most important. RRE and CoRRE are just too complicated. Hextile encoding can be simplified by not sending subrects at all, but either filling or sending raw data for the whole tile. Also, there could be a "dirty bitmap" that is 1/16 by 1/16 of the Display extent which has one bit for each hextile. When blitting to Display, a second blit could mark those tiles dirty ... -Bert-

1 0

Re: [FIX] close box flashing
by Bert Freudenberg 04 Apr '00

04 Apr '00

On Mon, 3 Apr 2000, John M McIntosh wrote: > A week back during a discussion about morphic damage updating I made the > comment about close box flashing. Jim Benson was kind enough to look but > didn't quite find the issue I saw. Early this week I looked and found the > problem, which is in: > > awaitMouseUpIn:whileMouseDownDo:whileMouseDownInsideDo:ifSucceed: Hey - that's MVC! We were talking about *Morphic* damage wich is handled quite differently. Not surprising we couldn't reproduce it. Anyway, thank's for fixing this. -Bert-

1 0

Re: VNC Server
by Bert Freudenberg 04 Apr '00

04 Apr '00

On Mon, 3 Apr 2000, Stephen Pair wrote: > My project for this weekend was to start implementing a VNC server for > Squeak. Cool! I thought about doing this myself - but then, Xvnc does all I want so there just wasn't enough motivation for me. It would be awesome for all the people that are forced to use Mac or Win :-) > 2. HandMorphs don't seem to be included in the Display bitmap (or maybe > they aren't their because I don't get a damage report for them)...how can I > pick up the HandMorphs? There are (at least?) two things to mention here: a) Squeak normally uses the hardware cursor. To make it always use it's own cursor drawing routines you can use World activeHand userInitials: ' '. Cursor blank show. b) The hand's damage is collected separately. Each hand (there can be more than one) saves a patch and restores it when moving - see PasteUpMorph>>displayWorld and the HandMorph's drawing protocol ~8-} OTOH I would much prefer intercepting BitBlt to Display, just like Dean said. It's more general. Good luck! -Bert-

1 0

Re: Graphical GUI building
by Marcel Weiher 04 Apr '00

04 Apr '00

Hi Chris, this looks very useful. Thanks! Marcel

1 0

Native GUI
by Karl Goiser 04 Apr '00

04 Apr '00

Hello, Please excuse me if this is an faq! Over the last couple of years I have picked up and had a brief look at Squeak and put it down again on several occasions. The main reason for putting down was because of the user interface. I'm a long time Mac user which, I guess, makes me some kind of UI Nazi, but I'd really far prefer to run Squeak using the Mac's user interface. Can you tell me if I'm stupid or that it might be possible? If it is, I'd certainly be willing to put some time in. Karl

1 0

RE: FixInterpreterPluginACG
by Raab, Andreas 04 Apr '00

04 Apr '00

Andrew, Just wondering - didn't the updates 1928/1931 work for you?! In particular, 1931 should make the #var:type: work when used within plugins itself. But I only tried it with the CodeGenerator - did you have problems with the TestCodeGenerator?! - Andreas > -----Original Message----- > From: Andrew C. Greenberg [mailto:werdna@gate.net] > Sent: Monday, April 03, 2000 7:28 PM > To: squeak(a)cs.uiuc.edu > Cc: recipient list not shown > Subject: FixInterpreterPluginACG > > > 'From Squeak2.8alpha of 14 January 2000 [latest update: #1919] on 3 > April 2000 at 10:12:34 pm'! > "Change Set: FixInterpreterPluginACG > Date: 3 April 2000 > Author: Andrew C. Greenberg > > Changes introduced in Update 1908 broke both plugin compilers, > causing interpreterProxy to be misdeclared in generated code as 'int > interpreterProxy', causing the generated plugins to be uncompilable > without turning off type checking. This changeset repairs that bug. > > The new var:type: absolutely does NOT work, presently generating > unlinkable procedures in the generated plugin. While this can be > repaired fairly easily with a modification of CCodeGenerator to strip > the parse tree (as is presently done with var:declareC), I question > the wisdom of introducing the feature with this selector. IMHO > var:type: will too easily be confused with var:as: > May I suggest, perhaps, var:declareCType:?"! >

2 1

RE: [ANN]New Morphic tutorial
by Stewart MacLean 04 Apr '00

04 Apr '00

Hi Juan, It is with interest that I have been reading your tutorial on and off today. It's a good one! In fact theres so much to it I suggest that you have an index into all your pages, so that I may easily recommence my reading. It would also enable the random selection of pages of interest. Many thanks and keep it up, Stewart > -----Original Message----- > From: Juan Manuel Vuletich [mailto:jmvuletich@sinectis.com.ar] > Sent: Sunday, April 02, 2000 2:14 PM > To: squeak(a)cs.uiuc.edu > Cc: recipient.list.not.shown; @cs.uiuc.edu > Subject: [ANN]New Morphic tutorial > > > Hi all > > I've just published a new Morphic Tutorial and Sample App. > The tutorial includes a description, some historic background > and a description of the different programming styles. The > Sample App is a Morphic, Client/Server Scrabble Game. Please > see it at > http://www.sinectis.com.ar/u/jmvuletich/studio.> html. Every > comment is welcome. > -- > Juan > Manuel Vuletich > jmvuletich(a)sinectis.com.ar > jvuletic(a)dc.uba.ar > http://www.sinectis.com.ar/u/jmvuletich > http://www.sinectis.com.ar/u/jmvuletich/studio.html >

1 0

FixInterpreterPluginACG
by Andrew C. Greenberg 04 Apr '00

04 Apr '00

'From Squeak2.8alpha of 14 January 2000 [latest update: #1919] on 3 April 2000 at 10:12:34 pm'! "Change Set: FixInterpreterPluginACG Date: 3 April 2000 Author: Andrew C. Greenberg Changes introduced in Update 1908 broke both plugin compilers, causing interpreterProxy to be misdeclared in generated code as 'int interpreterProxy', causing the generated plugins to be uncompilable without turning off type checking. This changeset repairs that bug. The new var:type: absolutely does NOT work, presently generating unlinkable procedures in the generated plugin. While this can be repaired fairly easily with a modification of CCodeGenerator to strip the parse tree (as is presently done with var:declareC), I question the wisdom of introducing the feature with this selector. IMHO var:type: will too easily be confused with var:as: May I suggest, perhaps, var:declareCType:?"!

1 0

Re: Toward version 2.8
by R. A. Harmon 04 Apr '00

04 Apr '00

At 10:10 AM 4/2/00 -0700, Dan Ingalls wrote: >We're converging on a plan for Squeak 2.8, so I thought I would let you know how it looks. [snip] I suggest also adding my implementation of the ANSI methods missing from Squeak. I can send you a change set of it. It currently includes the Fundamental, Valuable, Exception, Numeric, and Date And Time Protocols. The missing Collection, Stream, and FileStream Protocols can be added when finished. I am updating the Collection implementation for a previous version and will finish the Stream, and FileStream Protocols. -- Richard A. Harmon "The only good zombie is a dead zombie" harmonra(a)webname.com E. G. McCarthy Spencer, Iowa

3 2

Re: VNC Server
by Dean_Swan＠Mitel.COM 03 Apr '00

03 Apr '00

>3. Where can I hook in to make this work with MVC as well? InputSensor for all the mouse and keyboard stuff, and BitBlt>>copyBits for Display updates. You may be able to get away with just modifying InputSensor>>mousePoint, InputSensor>>keyboard, and InputSensor>>keyboardPeek, but you'll probably also need to intercept InputSensor>>primMouseButtons, since this is where all the keyboard modifiers and mouse buttons get polled from the host OS. What I've done in the past to intercept Display updates is to rename 'BitBlt>>copyBits' to 'BitBlt>>basicCopyBits', then substitute my own implementation of 'BitBlt>>copyBits', which ultimately calls 'BitBlt>>basicCopyBits'. In this case I imagine you'd want to do something like: copyBits (destForm == Display) ifTrue: [ "Do VNC Server Screen Refresh Logic" ]. self basicCopyBits. You do have to be careful when doing something like this, and I recommend that you manually adjust any changesets you distribute so that the changes to BitBlt are last in the file, or you may end up with a broken image when you try to file it in. When you break 'BitBlt>>copyBits', it gets rather difficult to tell what's going on, so I'd suggest writing lots of debug messages to a file, and always save your image BEFORE saving your new 'BitBlt>>copyBits', because if there are any problems in your new 'copyBits' method, you can pretty much count on having to kill the VM and start over (i.e. re-launch Squeak). I used this technique to implement cursor display in Smalltalk, for platforms where no hardware cursor display is available (i.e. WinCE, EPOC). That code works ok now, but I still have a few bugs to track down in it related to hiding the software cursor when crossing window boundaries, and when the cursor shape is changed, but I'd be happy to send you a copy if you want to look at it for reference. This may be pretty messy re-wiring of very low-level stuff, but if you do it right, then it will work for *BOTH* Morphic and MVC, without any need for a "Morphic-only" implementation. -Dean Swan dean_swan(a)mitel.com

1 0

Jump to page: