Squeak-dev August 1998

squeak-dev@lists.squeakfoundation.org

99 participants
188 discussions

Re: Problems building the Mac VM
by Jim Benson 11 Aug '98

11 Aug '98

O'NEEL Bruce wrote: > Hi, > I've poked at Squeak on my Mac at home, as well on the Solaris box > at work, and I'm not sure I see how to extract the C source and build > a VM from the distributed images. Could someone please point me to a > howto? > > Bruce, A brave undertaking. On a Macintosh, generating a VM is a multi step process. First generate the source code for the interpreter. Time millisecondsToRun: [ Interpreter translate: 'interp.c' doInlining: true. Smalltalk beep] ( this is taken from the Interpreter class ; category --translation ). Depending on the speed of your machine, this may take a while to run. This process generates the interpreter, memory manager, BitBlt, etc. Then generate the support code for the interpreter: InterpreterSupportCode writeMacSourceFiles This writes all the source files for the runtime/hardware support code to the current image directory. I usually place all of the generated files into a new folder before compiling. On the Mac, a stuff-it file called 'projectArchive.sit' is generated. Unstuffing reveals two CodeWarrior 8 project files, Squeak68K.proj and SqueakPPC.proj. (BTW, CodeWarrior is a C language "development environment"). Compile these puppies. Almost magically, a couple of new Squeak VMs are generated. VM generation for other machine platforms is usually a little different. Generally, the platform support ftp/web sit has the source code and the make files for that platform. As a sidebar, Interpreter generates itself through a Smalltalk to C translation mechanism, which is implemented through class CCodeGenerator. This is a useful tidbit should you decide to modify the VM yourself. Amaze your friends !!! Confuse your enemies !!! Build your own Squeak VM. jb

1 0

Re: Linux GGI?
by Tim Rowledge 11 Aug '98

11 Aug '98

On Tue 11 Aug, Chris Reuter wrote: > I thought I'd mention the Linux GGI project for those who aren't aware > of it. This is a scheme under development to move graphics drivers > into the Linux[1] kernal. This has several advantages, the relevant > one here being that non-suid program can now access the video > hardware, and without risk of crashing the computer. > > This is ideal for Squeak because it doesn't really need a windowing > system and X has a lot of overhead. This might make running Squeak on > 386 and 486 hardware feasible. When I did the DEC Itsy port a few weeeks ago we had to work around this same 'problem'. Itsy has no X-Windows (thank goodness) and we just used the framebuffer device in what I believe is 'the old fashioned way', i.e. /* open the framebuffer device */ if ((fb_d = open("/dev/fb", O_RDWR)) < 0) { perror("Error opening /dev/fb\n"); return (unsigned char*)0; } /* setup the frame buffer */ if ((k=ioctl(fb_d, FB_LCD_PARAMS, &fbLcdParams)) != 0) { perror("Error with /dev/fb ioctl FB_LCD_PARAMS call"); return (unsigned char*)0; } I don't know if Linux generically provides a framebufer device, or if this is an Itsy specific change, but I do remember using a very similar technique some years ago on RiscIX. Eliot's BrouHaHa VM also did something very similar for the various unices it ran on. tim -- Useful random insult:- Hears more lyrics on records when they're played backwards. Tim Rowledge: rowledge(a)interval.com (w) +1 (650) 842-6110 (w) tim(a)sumeru.stanford.edu (h) <http://sumeru.stanford.edu/tim>

1 0

Problems building the Mac VM
by stp＠limbo.create.ucsb.edu 11 Aug '98

11 Aug '98

Hello all, I seem to be having a problem rebuilding the Mac 2.1 VM. It compiles and links just fine, but then I get an unrecoverable error on the first file access. Has anyone else seen this behavior? I'm using CodeWarrior 11 Gold. ....any advice appreciated... stp _ Stephen Travis Pope _ Center for Research in Electronic Art Technology (CREATE) _ Department of Music, Univ. of California, Santa Barbara (UCSB) _ stp(a)create.ucsb.edu, http://www.create.ucsb.edu/~stp/

3 3

Pass thru proxy server
by Jim Benson 10 Aug '98

10 Aug '98

Hi, I am trying to build a "pass through" web proxy server using the pluggable web server in Squeak. I want to : Browser --> Squeak Server --> Network connection where the Squeak Server is invisible both to the browser and to the network connection other than the address redirection. >From the browser, I just add a HTTP Server Action that passes the request on to the real network. The reason that I want Squeak in the middle is to collect html /text pages as they pass through. I then add these pages to a database that I can query. This is useful for doing web based research, where you typically gather info and articles rather haphazardly. With Squeaks help, you would be able to browse as you normally would, Squeak gathers everything you surf, and later you can do some text /information mining using a more sophisticated interface. So it sounds pretty simple, my HTTP Server Action is dispatched on a scheme type of 'http'. HTTP Server Action does a GET using a MIME of 'application/octet', and in general, pretty much everything gets through. EXCEPT -- when the network connection wants something in return. Cookies are a good example of this, as are Authentication scripts. So my question is -- How do I make PWS transparent bidirectionally? In other words, how do I make the Squeak Server invisible to both the browser and the network other than the address relocation? Again humbled by seemingly simple things, jim

1 0

Re: Disabling Palm OS mem protection (was: Status of Palm III port?)
by wirth＠almaden.ibm.com 09 Aug '98

09 Aug '98

Maurice Rabb responded: ......Have you been following the thread "Smalltalk in A PDA" on c.l.s? Andrew Brault has been working on a really interesting Smalltalk-- for the Pilot. It is not Squeak, but it is very very small. He mentions that he disables memory protection to get around. .... > I _think_ I have most of the memory problems worked out. It all > seems to boil down to turning memory protection off with > MemSemaphoreReserve() (gack!) .... > Andrew Brault <ajb(a)tiac.net> ========================= Maurice, Thanks!!! I took a quick look in my Palm programming library (about 4,000 pages of Palm programming docs, in Acrobat PDF and indexed) and found the following three references to MemSemaphoreReserve: In Darrin Massena's Pilot Hack Tutorial (http://www.massena.com/darrin/pilot/PilotHackTutorial.htm): ====================================================== Very interesting! The first 32K of the Pilot's address space is RAM that is duplicated (simultaneously mapped to) the range $1000000-$10007fff. The memory from $0000000-$00007fff can be read from, written to, and executed from but the memory starting at $10000000 can only be read or executed -- no writing without gaining permission from an API (I later discovered) called MemSemaphoreReserve. The Pilot appears to support at least a primitive form of hardware memory protection which is used to separate "Dynamic" (Palm's term) from "Storage" RAM and keep rogue programs from trashing persistent system data structures, user data, and other programs accidentally. This is not real data security, just a very sensible precaution that should protect from the majority of random software happenings. In the GNU Pilot Programming FAQ (http://www.geocities.com/SiliconValley/Horizon/5249/GNU_Pilot_SDK.txt): ====================================================== 7.3.2. Can I use a Database to simulate large memory chunks? Yes you can. Take a look at the source code for Wes Cherry's 2-bits-per-pixel bitmap utilities at <URL:http://www.scumby.com/scumbysoft/pilot/win2/>. The function _Win2DoAllocScreen() in the file win2.c uses the API function MemSemaphoreReserve() to accomplish this trick. Note: This code contains the comment: | Note that you must call MemSemaphoreReserve(1) to gain write-access to the chunk | (call MemSemaphoreRelease(1) to restore protection) And in the official Palm OS 3.0 docs, Developing Palm OS 3.0 Applications, Part III: Memory and Communications Management ====================================================== MemSemaphoreReserve Prototype Err MemSemaphoreReserve (Boolean writeAccess) WARMING(sic :-): This function for use by system software only. ====================================================== Guess I should have gone to the net to ask this question earlier :-) Thanks for the help. Now the question is when I'll have time to put any work in on the port of Squeak to the Palm OS (since I'm working on another port now). Mike Wirth

1 0

About Squeak on OS/2
by vuletich＠ar.ibm.com 08 Aug '98

08 Aug '98

To: squeak(a)cs.uiuc.edu cc: From: Hi Boris, I apologize for speaking about the performance of your work without actually trying it. I never saw Squeak running in UNIX, but I thought that the speed of a game oriented api like DIVE or DirectX should be better than that of XWindows. I also think that DirectX could help get a faster Squeak on low end Windows systems. Again, I'm not really shure about it. I will first add sound and MIDI support (I will then start to work on sound and music modelling, maybe I'll port Siren to OS/2). But I don't know much about TCP/IP, and if you want to help me with that I'll be very glad. Could you refresh me where can I download Cheese? Does it still need XFree86? (I never used the first releases 'cause I don't have it). Btw, could you finally get something working on the AS/400? Thank for your comments. Juan Manuel Vuletich jvuletic(a)dc.uba.ar

2 1

Re: (Status of) Palm III port?
by Mike Klein 07 Aug '98

07 Aug '98

On Fri, 7 Aug 1998 wirth(a)almaden.ibm.com wrote: > Any ideas? What percentage of bytecodes executed are writes to objects (as > opposed to the stack)? FWIW, I did some quick tests in VW3.0 (I'm still not quite up on building squeak VM's, ain't got no C compiler, but the numbers should be indicative). Compiling and evaluating 3+4 has 4% (284/6966) instVar stores. Decompiling and pretty-printing Object has 2.7% (40620/1509278) instVar stores. I'm curious what the numbers look like for a typical day of Squeak development. -- Mike Klein

1 0

Squeaking the Web
by Alejandro F. Reimondo 07 Aug '98

07 Aug '98

Folks, I'm very happy because I can implement the Squeak Virtual machine as an ActiveX component. Now we can embedd a Squeak image in a WEB page! It is very important, because now we can write active content for the web in Squeak. We can load a Squeak image in a web page, perform a filein from a remote server, save the image in the local disk... Provably in the near future we (the people of sugar) can show our miniatures working in the web, and people will use browsers to see the code and inspect the living objects directly from the web page! It is really exiting. You can visit my pages on Smalltalk in the Web at http://www.sugarWeb.com/Web/Squeak/readme.htm or http://www.sugarWeb.com/Web/index.htm (in spanish) Please visit the page, and see the examples. I will be waiting for your comments. Ale. Alejandro F. Reimondo Feel free to visit Smalltalk User Group of Argentina (SUGAR) at http://www.sugarWeb.com

4 3

Re: [Squeak] Stupid question on messages sent to objects of same type
by Tim Olson 07 Aug '98

07 Aug '98

>However, when I hit the multiplication in the mandelbrotStepsToInfinity: >method (which sends a message to a temporary variable of ComplexNumber), the >debugger kicks out a "does not understand message", when trying to multiply >component parts of the ComplexNumber. The problem is in the mandelbrotStepsToInfinity: method; there you create a new Complex number with "new", rather than "new:imaginary:", thus the instance variables "real" and "imaginary" aren't initialized. You should either cause "new" to initialize these variables, or change the creation code to use "new:imaginary:". (the same problem exists in the other arithmetic methods). -- tim

1 0

RE: Squeaking the Web
by Ranjan Bagchi 07 Aug '98

07 Aug '98

[Totally off topic] Let's not forget that interpretations for URL's is up to the web server: If a more intelligent naming scheme is desired the server's free to implement it. What we're talking about [in the human-readable sense] is the naive default implementation that most servers put in and relies upon a file system. I'm doing server-side java stuff professionally and one of the cool things about servlets is that they can apply some knowledge about how to interpret URLs to deliver the right pages. URL's like http://foo.com/servlet/SmartServlet/SomethingUseful can be totally context-dependent on what pages are served up. -rj -----Original Message----- From: Mark Guzdial [SMTP:guzdial@cc.gatech.edu] Sent: Wednesday, August 05, 1998 7:28 AM To: squeak(a)cs.uiuc.edu Subject: Re: Squeaking the Web >Lex Spoon writes: > > > > 2) Links are so easily broken. Maybe *all* links should go > > through a level of indirection on a stable server? Do links *really* > > need to be all that human readable? > > > >Actually, this was the origional idea. URLs were a quick hack until >URNs worked. URNs were going to work a bit like DNS does with a >distributed database so you would not have broken links and give you >that extra level of indirection. Hmm, this was '94 or so :-) On the other hand, I've read some analysis (armchair quarterbacks, I admit :-) that suggest that the Web succeeded where previous hypertext efforts failed (e.g., Intermedia, Xanadu) explicitly because URL's were one-direction only. All previous efforts made sure that links couldn't break, and this analysis claims that that extra bit of engineering is what pushed the system over the edge to overly-large and less-adoptable. Mark -------------------------- Mark Guzdial : Georgia Tech : College of Computing : Atlanta, GA 30332-0280 (404) 894-5618 : Fax (404) 894-0673 : guzdial(a)cc.gatech.edu http://www.cc.gatech.edu/gvu/people/Faculty/Mark.Guzdial.html

4 3

Jump to page: