Squeak-dev December 2004

squeak-dev@lists.squeakfoundation.org

194 participants
491 discussions

gui objects for inputing integers
by Ralph Boland 03 Dec '04

03 Dec '04

I need to build a pop up window where I have (amoung other things) a set of boxes with which to input integers. Can someone tell me which GUI object I should use for these boxes? I also want to be able move the cursor to a given box by clicking the mouse when the cursor is in the box or by typing the tab character to have the cursor cycle through the boxes, one box per tab. If anyone can point me to code that I can study/steal from in order to implement my window that would be most appreciated. Thanks in advance. Ralph Boland __________________________________ Do you Yahoo!? Meet the all-new My Yahoo! - Try it today! http://my.yahoo.com

3 2

RE: [Q] Reading little endian DOUBLES on big endian platforms
by Baveco, Hans 03 Dec '04

03 Dec '04

Whow, thanks! Is it really that simple: a float being just these bytes in the right order? Could this method be added to ByteArray, in basic Squeak? Hans From: "Andreas Raab" <andreas.raab@g...> Date: Fri Dec 3, 2004 11:09 am Subject: Re: [Q] Reading little endian DOUBLES on big endian platforms doubleAt: index bigEndian: bool | w1 w2 dbl | w1 := self unsignedLongAt: index bigEndian: bool. w2 := self unsignedLongAt: index+4 bigEndian: bool dbl := Float new: 2. bool ifTrue:[ dbl basicAt: 1 put: w1. dbl basicAt: 2 put: w2. ] ifFalse:[ dbl basicAt: 1 put: w2. dbl basicAt: 2 put: w1. ]. ^dbl -----Original Message----- From: Baveco, Hans Sent: Friday, December 03, 2004 11:18 AM To: The general-purpose Squeak developers list Subject: RE: [Q] Reading little endian DOUBLES on big endian platforms I am too stupid to be able to write this method. Until a #doubleAt:bigEndian: methods pops up I will use the workaround of swapping the bytes on big endian platforms and using #doubleAt: on the swapped bytes. It's not elegant and it may slow down the reading for Mac users, but it should work... Hans No trick. The method doesn't exist. If you need it you'll have to write your own. Cheers, - Andreas ----- Original Message ----- From: "Baveco, Hans" <Hans.Baveco@w...> To: "The general-purpose Squeak developers list" <squeak-dev@l...> Sent: Friday, December 03, 2004 9:39 AM Subject: [Q] Reading little endian DOUBLES on big endian platforms In my Shapes package I make extensive use of the ByteArray>>doubleAt: method (in the *FFI category), to read little-endian DOUBLES from the contents of a binary file stream. This #doubleAt: is part of the FFIPlugin, and maybe not a good general thing to rely on (can it be absent in some Squeak distributions)? On windows it works fine, but on Mac not (of course). There doesn't seem to be a doubleAt:bigEndian: method available that allows me to define the endianness, like there is for integers (longAt:bigEndian: & shortAt:bigEndian:). What is the trick here? Hans

2 1

RE: [BUG][FIX]numbers in scientific notation with capital E are nothandled correctly (not good)
by Baveco, Hans 03 Dec '04

03 Dec '04

OK, that's clear. It felt already too goo to be true... (I never think hexadecimal). Funny how much knowledge and experience there is behind many methods in basic squeak/smalltalk. A pity though that so little is documented in/near the source :-(. Wrt my problem of reading the uppercase exponent: the method Number>>readRemainderOf:from:base:withSign: is executed after having read the 'front' part of the number. Do we not know already whether the number is a decimal number or not? In other words, would it be possible to limit this check (peekChar = $E) to decimal numbers only? I might overlook complications again, though.. but it is just not an attractive option, having to scan every possible input file for occurrences of $E. Thanks a lot for your informative reply, Hans -----Original Message----- From: Bert Freudenberg [mailto:bert@impara.de] Sent: Wed 12/1/2004 7:33 PM To: The general-purpose Squeak developers list Cc: Subject: [BUG][FIX]numbers in scientific notation with capital E are nothandled correctly (not good) Am 01.12.2004 um 10:15 schrieb Baveco, Hans: > I have often been bit by this feature of squeak. '1234.123E-5' > asNumber produces 1234.123 because the exponent part is ignored. Some > numeric software produces output in this format, and reading datafiles > from such programs may lead to very hard to trace anomalies > (especially when scientific notation is only used for some of the > numbers). A simple one-liner change in > Number>>readRemainderOf:from:base:withSign: seems sufficient to handle > capital E, see attached fileout. > > I changed "peekChar = $e | (peekChar = $d) | (peekChar = $q)" > > to "((((peekChar == $e) or: [peekChar == $E]) or: [peekChar == $d]) > or: [peekChar == $q])" > > (I compensated for the extra check by replacing the bars by blocks). > > I can't imagine there are any unwelcome side-effects associated with > this change? It changes the language, as this method is also used for reading literal numbers. I've been down that road before and was educated by Dan Ingalls that things are as they are for a reason. The reason in this case is that uppercase letters are reserved for digits (think hexadecimal). You can enter literal floats up to base 36 in Smalltalk: 16rABC.DEF So, just convert your numbers to lowercase before reading them in. - Bert -

2 1

[ANN] Shapes 1.3 also for big endian platforms
by Baveco, Hans 03 Dec '04

03 Dec '04

Thanks to Andreas' doubleAt:bigEndian: I could make the ESRI shapefile reading/viewing package on SqueakMap work on both little and big endian platforms. Hans

1 0

RE: [Q] Reading little endian DOUBLES on big endian platforms
by Baveco, Hans 03 Dec '04

03 Dec '04

I am too stupid to be able to write this method. Until a #doubleAt:bigEndian: methods pops up I will use the workaround of swapping the bytes on big endian platforms and using #doubleAt: on the swapped bytes. It's not elegant and it may slow down the reading for Mac users, but it should work... Hans No trick. The method doesn't exist. If you need it you'll have to write your own. Cheers, - Andreas ----- Original Message ----- From: "Baveco, Hans" <Hans.Baveco@w...> To: "The general-purpose Squeak developers list" <squeak-dev@l...> Sent: Friday, December 03, 2004 9:39 AM Subject: [Q] Reading little endian DOUBLES on big endian platforms In my Shapes package I make extensive use of the ByteArray>>doubleAt: method (in the *FFI category), to read little-endian DOUBLES from the contents of a binary file stream. This #doubleAt: is part of the FFIPlugin, and maybe not a good general thing to rely on (can it be absent in some Squeak distributions)? On windows it works fine, but on Mac not (of course). There doesn't seem to be a doubleAt:bigEndian: method available that allows me to define the endianness, like there is for integers (longAt:bigEndian: & shortAt:bigEndian:). What is the trick here? Hans

2 1

Bug 0583 - likely emergency issue
by Tim Rowledge 03 Dec '04

03 Dec '04

In a 6461 image, it is possible to load VMMaker from SM. In a 6475 updated image it causes a vm-fatal crash. Of the changes from 6461 to 6475 the only ones that look even faintly possible culprits are the Behaviour changes. Could someone with a faster machine and network than I have please try out some combinatorial tests of this? Try reverting some of the Behaviour changes and then loading VMMaker off SM to see if anything looks guilty. tim -- Tim Rowledge, tim(a)sumeru.stanford.edu, http://sumeru.stanford.edu/tim All the simple programs have been written, and all the good names taken.

3 5

[UPDATES] 3.8 gamma and 1 update
by Michael Rueger 03 Dec '04

03 Dec '04

Hi all, 3.8 has now been moved to gamma and one more fix. If there are fixes out there that are still missing or urgent bugs that you think should be fixed before final, now is the time to speak up. Michael ------------ Change Set: startUp-bf Date: 23 November 2004 Author: Bert Freudenberg Fix auto start if more than one launcher is active

7 11

[ANN] BFAV 2.30
by Frank Shearar 03 Dec '04

03 Dec '04

This announcement heralds the release of BFAV2 2.30: SqueakMap now has published releases for both Squeak 3.8 and 3.9 alpha. If you tried to install BFAV today and you get an MNU about something not being able understand #Undeclared, then remove the BFAV-Installer-2.30.st from your SqueakMap cache (it should be in ./sm/cache/packages/uso29ityq9g8qmdlwolxhv/12) and try again - I only discovered a syntax error in the .st after I uploaded the file to SqueakMap, and so I deleted it and reuploaded the file). A summary of changes (listed using the Monticello version numbers): 179: Based on Tom's review, this commits Karl Ramberg's BFAVAttachmentFix 178: Added German Morales's progress reporter when loading posts and Frank Shearar's ensure block around updateListing 177: Second release moving us toward having ability to transfer post. This one adds a DualPatchArchiveClient (a la DualChangeSorter), so that eventually we will be able to drag posts from one side (group) to the other). Adds a set of (uninspired) tests for the DualPAC. To make this work I made both PAC and DualPAC be subclasses of model rather than SystemWindow. This makes them like the DRNP. 176: Implemented workaround for [BUG][BFAV]Loading updates failed Archive ID: 24104 and commented out testSend messages that were the original source of problem. Implemented enhancements for [ENH] [BFAV]PatchArchiveRegexpFix Archive ID: 24163 175: Added the first pieces of logic for transfering posts from one group to another into DRNP. See initializeUsingTargetPost: anArchivePost sourcePost:. Upgraded tests by: adding a DNRPTransferTest, abstracting functionality into MorphicTestCase and removing ArchiveResource (which incorrectly used classVars). 174: Adds a Bug tab to the BFAV where all bug (without fixes) reports reside. The Unreviewed and Reviewed tab shows only fixes and enhancements. 173: Fixes implementation of makeUnclosable/makeClosable to handle send and to do in one place 172: Frank Shearar wrote: Now that BFAV uses SystemWindows for the DRNP, if you click the close button you don't get a confirmation dialog, and the window just closes. Karl Ramberg wrote: Send a 'self makeUnclosable' to the window at some point when the DRNP is initialized. This release does that. The side effect is that there is no pink x icon on the far left. 171: This is just a refactoring release. I have cleaned up DRNP and PAC in preparation for adding the ability to split post groups and to merge post groups. I removed a considerable number of dead methods (and tests) and focused on getting the tests for PA and PAC to be more realistic. Merged in changes from 169 and 170 since this work had been based on 168 170: The DecentReviewerNotePad didn't correctly open on Changesets and ContextParts (using #openOnChangeset: and #openOnContextPart:). 169: This change commits Samuel Tardieu's ReviewerPost1 changeset, which adds a new header (X-Squeak-Version) to the mails sent by BFAV. 168: Created first version of the reopen capability. Add a new harvest button (Open) which resets any prior approve, update or close events. Rule for categorizing posts, following Doug's suggestion, allows for multiple close, open, close, open events by taking the last relevant event as the groups status. I've done the simplest thing that could possibly work: added one new harvest status with a mask of 16 and a tag of [open]. If you want to back up one step from say update to approve, you've got to open, and then approve. I think this will work. Has temporary code to work w/o server side changes (until Ken makes his changes.) In the process eliminated few unused method, reorged categories and added more class comments. 167: Fixes [BUG] BFAV Post both reviewed and Unreviewed reported by Frank Caggiano on 10 April 2004 as Archive ID: 21795. Fix follows ken(a)kencausey.com suggestion that "more than one path in the decision network for partitioning the messages" The new implementation ensures a partition and sets as unreviewed any group not otherwise categorized. 166: This fixes the problem where clicking on a SystemWindow didnot bring it to the forground, if DRNP happened to be opened. As diagnosed by Ned, this was because DRNP was a morph rather than a System Window. This fixes the problem using Ned's fix in most places. As always, don't hesitate to let Tom or I know if something's not right with BFAV2. Enjoy! frank ******************************************************************* This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. Any views or opinions expressed are solely those of the author and do not necessarily represent RNID policy. If you are not the intended recipient you are advised that any use, dissemination, forwarding, printing or copying of this email is strictly prohibited. If you have received this email in error please notify the RNID Helpdesk by telephone on: +44 (0) 207 296 8282. The Royal National Institute for Deaf People Registered Office 19*23 Featherstone Street London EC1Y 8SL No. 454169 (England) Registered Charity No. 207720 ********************************************************************

4 3

[Q] Reading little endian DOUBLES on big endian platforms
by Baveco, Hans 03 Dec '04

03 Dec '04

In my Shapes package I make extensive use of the ByteArray>>doubleAt: method (in the *FFI category), to read little-endian DOUBLES from the contents of a binary file stream. This #doubleAt: is part of the FFIPlugin, and maybe not a good general thing to rely on (can it be absent in some Squeak distributions)? On windows it works fine, but on Mac not (of course). There doesn't seem to be a doubleAt:bigEndian: method available that allows me to define the endianness, like there is for integers (longAt:bigEndian: & shortAt:bigEndian:). What is the trick here? Hans

2 1

[ENH] bench (was: Object turnover.)
by Chris Muller 03 Dec '04

03 Dec '04

> Should I worry about this object turnover, or is the system designed > to handle this high object turnover? Just trying to get my head > around some of the issues of efficient squeak code.. The wonderful thing is, analyzing performance is a breeze in Squeak! Check out this changeset, it's just one method and is my favorite "quick benchmarking" method. [ Array new: 100 ] bench ----> '622021.595680864 per second.' [ Association allInstVarNames ] bench ----> '77790.0419916017 per second.' [ "as fast as it can go" ] bench ----> '3.446075384923015e6 per second.' It is much more useful and convenient than Time millisecondsToRun: - less typing, you can bench anything with one word, "bench" - it always runs in about five seconds, not 1 second or 5 minutes. You get a reasonable sample and a reasonable wait, every time, with NO guesswork. - it actually tells you how fast it is (executions / sec) instead of how long it took providing a consistent measurement for different things benched over time. It's so easy I find myself frequently using it while coding a method and I don't even trouble myself opening a workspace; I just type "bench" right there where I'm at. Cheers, Chris

3 2

← Newer
1
...
41
42
43
44
45
46
47
...
50
Older →

Jump to page: