Squeak-dev November 1999

squeak-dev@lists.squeakfoundation.org

164 participants
545 discussions

RE: Byte order and ByteArray conversions [SUMMARY]
by Pennell, David 03 Nov '99

03 Nov '99

If you expect to use isLittleEndian a lot, it might be worthwhile to compute this during image startup and cache the results in a class variable. Look for selectors containing "startUp" for examples of how to run code at image startup. -david [snip] > isLittleEndian > "Answer true if the machine on which we are executing uses little > endian byte order conventions." > > ^ ((ReadStream on: (ByteArray newFrom: #(0 0 4 210))) > nextNumber: 4) == 1234 [snip]

1 0

Re: B3D busted on Linux?
by Raab, Andreas 03 Nov '99

03 Nov '99

Just to let you know: There is a tiny little definition when compiling the plugin that should do the trick: You *MUST* define LSB_FIRST if you're compiling on a little endian platform (the compiler defaults to MSB_FIRST because I can't get stupid CodeWarrior to have global #define for this kind of stuff - argh!!!). Everything should be okay then. Andreas PS. I know I should have taken the HAS_MSB_FIRST/HAS_LSB_FIRST definition from sq.h and I'll do that in the future. For now you'll just have to define it manually when 'make'ing the plugin. > Duane, > I got the same problem with my 3D facial animation stuff. I have a texture > for the face, but the colors look wrong (I get a bluish face). It's > strange. If I display the texture on the screen doing > texture displayAt: 100@100 > it looks correct, but in the face it's wrong. > However, the bunny's texture (not the ground) seems to display the same in > the bunny's body than in the screen... I don't know why. > Let me know if you find what's happening... > Luciano.- >> Hey all - >> >> We're trying to diagnose a problem with B3D on Squeak 2.5 and 2.6 under >> Linux, and was hoping that someone out there could confirm what we're >> seeing. >> >> The simplest thing to check is "Play With Me - 7". On our machine, the >> ground is a pale blue instead of green. In general, the entire bunny >> image >> has a bluish cast. Also, in the B3DSceneMorph, the cube's colors are >> washed >> out. >> >> We're occasionally getting Segmentation Faults. >> >> On the other hand, we're seeing some seemingly correct behaviors under >> Linux that we're not seeing on the Mac. In the B3DSceneMorph, the >> rotating >> cube is supposed to have 50% translucent faces, which they do under >> Linux, >> though pale as noted above. On the Mac, they are rendered totally >> opaque. >> >> We cannot detect problems in other subsystems - color is rendered >> properly >> in the rest of Balloon (or at least consistently across platforms) and >> BitBlt. >> >> We checked all these on a Linux 2.6 build downloaded just today. We have >> reproduced these problems under 2.5 also. >> >> If you can confirm one way of the other if you're seeing the same >> behavior, >> please let me know. >> >> Thanks! >> >> -- Duane >> >> >> >> >>

1 0

Re: B3D busted on Linux?
by Luciano Notarfrancesco 03 Nov '99

03 Nov '99

Duane, I got the same problem with my 3D facial animation stuff. I have a texture for the face, but the colors look wrong (I get a bluish face). It's strange. If I display the texture on the screen doing texture displayAt: 100@100 it looks correct, but in the face it's wrong. However, the bunny's texture (not the ground) seems to display the same in the bunny's body than in the screen... I don't know why. Let me know if you find what's happening... Luciano.- > Hey all - > > We're trying to diagnose a problem with B3D on Squeak 2.5 and 2.6 under > Linux, and was hoping that someone out there could confirm what we're > seeing. > > The simplest thing to check is "Play With Me - 7". On our machine, the > ground is a pale blue instead of green. In general, the entire bunny image > has a bluish cast. Also, in the B3DSceneMorph, the cube's colors are washed > out. > > We're occasionally getting Segmentation Faults. > > On the other hand, we're seeing some seemingly correct behaviors under > Linux that we're not seeing on the Mac. In the B3DSceneMorph, the rotating > cube is supposed to have 50% translucent faces, which they do under Linux, > though pale as noted above. On the Mac, they are rendered totally opaque. > > We cannot detect problems in other subsystems - color is rendered properly > in the rest of Balloon (or at least consistently across platforms) and > BitBlt. > > We checked all these on a Linux 2.6 build downloaded just today. We have > reproduced these problems under 2.5 also. > > If you can confirm one way of the other if you're seeing the same behavior, > please let me know. > > Thanks! > > -- Duane > > > > >

2 1

RE: [ENH] Scamper content type handling
by John Duncan 03 Nov '99

03 Nov '99

Yeah, Lex, you're on the right track. Personally, I'm not that worried about Scamper yet, since I expect it to be very much different very soon. If anyone is interested, here's my take on the situation: > > > 'text/html; charset=iso-8859-1' ^^^^^^^^^^^^^^^^^^ is an example of some sort of annotation, perhaps introduced in a more recent version of MIME or HTTP. What I expect is that the header processor in HTTPSocket currently treats this line as always an 'x/x' string, though I can imagine many other things going in there: 'text/html; charset=iso-8859-1; language=en-us' etc. So, if I'm right, then MimeDocument should keep a dictionary or list of annotations and HTTPSocket has to be changed to use it. That's more than one fix, and Scamper currently isn't very smart, so it's good enough to just filter out the rest of it in the content string. Plus, it'll fit into the 640k of memory we set aside for applications. :) -John > -----Original Message----- > From: Lex Spoon [mailto:lex@cc.gatech.edu] > Sent: Tuesday, November 02, 1999 11:34 AM > To: squeak(a)cs.uiuc.edu > Subject: Re: [ENH] Scamper content type handling > > > I think there is a contentType method in HTTPSocket which is > what is used when the MimeDocument is created. Put your > fix in there, > and it would affect everything that uses MimeDocuments. > > If that doesn't work, peek into getDocument:etc and postDocument:etc > methods on the class side of HTTPSocket and see how where > the documents content-type comes form. Fix them to use the > contentType method :) > > Lex > > > > > > One note: This is in the quick-fix category. If there is > a standard > > method of dealing with annotations, it should be in the > MimeDocument > > class, not Scamper. I just don't have the time to muck around with > > this stuff yet. > > > > -John > > > > > -----Original Message----- > > > From: John Duncan [mailto:jddst19+@pitt.edu] > > > Sent: Tuesday, November 02, 1999 9:59 AM > > > To: Squeak > > > Subject: [ENH] Scamper content type handling > > > > > > > > > > > > Some locations, notably the W3C, return content types like this: > > > > > > 'text/html; charset=iso-8859-1' > > > > > > This used to confuse Scamper [2.6], who, after this cs, now > > > looks for > > > the 'text/html' substring. Eventually, I'm sure, we'll > make use of > > > charset annotations for the multilingual stuff. > > > > > > -John > > > > > > -- > > > "In any event, once Robert Craft forged the > > > Stravinsky-Schoenberg axis > > > in the 1950s and the eclecticism of the 1960s > alleviated the austere > > > serialism of the previous decade, the futures market in > Hindemithian > > > repose was struck by panic selling." > > > -Glenn Gould > > > > > > > > > > > >

1 0

[ANN] Virtual machines mailing list
by Dwight Hughes 02 Nov '99

02 Nov '99

The following announcement was made on comp.compilers today. I hope it will be of interest. -- Dwight -------------------------------------------------------------------- (beginning of original message) Subject: Virtual machines mailing list From: John R Levine <johnl(a)iecc.com> Date: 1999/11/02 Newsgroups: comp.compilers I got several requests to join a virtual machines mailing list, so I set one up. To subscribe, send a message to majordomo(a)iecc.com containing: subscribe virtmach or subscribe virtmach-digest and respond to the confirmation message. Once subscribed, send contributions to virtmach(a)iecc.com. The list is unmoderated but I will kick out anyone who's persistently disruptive, the same as the rest of the lists I run here. If someone wants to collect a FAQ at some point, I can put it on a web site. Regards, John Levine, comp.compilers moderator, johnl(a)iecc.com, http://iecc.com/compilers (end of original message) ---------------------------------------------------------------------

1 0

Enhan. For SUnit 2.6
by R. A. Harmon 02 Nov '99

02 Nov '99

Hi Robert Hirschfeld: Nice job on the SUnit 2.6 update. I find the following assertion useful: !TestCase methodsFor: 'asserting' stamp: 'RAH 10/23/1999 07:01'! should: aBlock raise: anException #chgRAH. "99/10/21 Use ANSI exceptions." aBlock on: anException do: [ :except | ^self]. TestFailure signal: 'Should have raised ' , anException printString ! ! I lifted it directly from a VSE port of the testing framework I got off the web. I'm not sure it works like it should in Squeak. I'm still struggling to get a handle on exception handling. Please let me know if it's flawed. Bye, -- Richard A. Harmon "The only good zombie is a dead zombie" harmonra(a)webname.com E. G. McCarthy Spencer, Iowa

1 0

B3D busted on Linux?
by Duane Maxwell 02 Nov '99

02 Nov '99

Hey all - We're trying to diagnose a problem with B3D on Squeak 2.5 and 2.6 under Linux, and was hoping that someone out there could confirm what we're seeing. The simplest thing to check is "Play With Me - 7". On our machine, the ground is a pale blue instead of green. In general, the entire bunny image has a bluish cast. Also, in the B3DSceneMorph, the cube's colors are washed out. We're occasionally getting Segmentation Faults. On the other hand, we're seeing some seemingly correct behaviors under Linux that we're not seeing on the Mac. In the B3DSceneMorph, the rotating cube is supposed to have 50% translucent faces, which they do under Linux, though pale as noted above. On the Mac, they are rendered totally opaque. We cannot detect problems in other subsystems - color is rendered properly in the rest of Balloon (or at least consistently across platforms) and BitBlt. We checked all these on a Linux 2.6 build downloaded just today. We have reproduced these problems under 2.5 also. If you can confirm one way of the other if you're seeing the same behavior, please let me know. Thanks! -- Duane

1 0

Menus and Balloon Help
by David N. Smith (IBM) 02 Nov '99

02 Nov '99

I'm finally actually programming in Morphic instead of just wishing I had the time, and will be probably asking lots of stupid questions for a while. I've built menus with balloon help but the balloons only appear when the menu is traversed with no key down. That is, when an Option-Click (on the Mac) pops up the menu and one keeps holding the button down to select there is no balloon help. If the menu is Option-Clicked and forced to stay up then one can get balloon help by moving the pointer over the menu. I've found a way to do selection with Option-Click-Drag, or keep the menu up with Option-Click and see the help, but it requires another menu item to dismiss the menu if no other selection is made. (Further, the little menus can easily get 'lost' behind windows so I have had to keep track of each menu so that they can be closed later if the user doesn't do it.) All of this works, but seems a lot more effort that I had hoped. Is there any way to get balloon help to appear during an Option-Click-Drag operation? Thanks! Dave _______________________________ David N. Smith IBM T J Watson Research Center Hawthorne, NY _______________________________ Any opinions or recommendations herein are those of the author and not of his employer.

1 0

[ENH] Scamper content type handling
by John Duncan 02 Nov '99

02 Nov '99

------=_NextPart_000_0011_01BF2518.F1B188D0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Some locations, notably the W3C, return content types like this: 'text/html; charset=iso-8859-1' This used to confuse Scamper [2.6], who, after this cs, now looks for the 'text/html' substring. Eventually, I'm sure, we'll make use of charset annotations for the multilingual stuff. -John -- "In any event, once Robert Craft forged the Stravinsky-Schoenberg axis in the 1950s and the eclecticism of the 1960s alleviated the austere serialism of the previous decade, the futures market in Hindemithian repose was struck by panic selling." -Glenn Gould ------=_NextPart_000_0011_01BF2518.F1B188D0 Content-Type: text/plain; name="ScamperContentTypeChange.2Nove9" ; x-mac-type="65417070" ; x-mac-creator="43534F6D" Content-Disposition: attachment; filename="ScamperContentTypeChange.2Nove9" Content-Transfer-Encoding: imap_stub 0,755,2,1973,0, ------=_NextPart_000_0011_01BF2518.F1B188D0--

1 1

Byte order and ByteArray conversions
by David T. Lewis 02 Nov '99

02 Nov '99

I'm sure I'm missing something obvious, but I'd like to be able to convert integers to and from byte arrays in native machine byte ordering, without resorting to a new primitive. 1) Does Squeak know the byte order convention of the hardware it's running on? I can't find any method to query for this. 2) Class ByteArray knows how to convert itself into integers, given that you can tell it the answer to question 1. Is there any existing method for converting an integer into a four byte array? Thanks.

2 2

Jump to page: