December 2010 - Oberon - mailman3.inf.ethz.ch

PlugIn Oberon for Windows
by Alexander Iljin 04 Feb '11

04 Feb '11

Hello! I'm trying to run ETH PlugIn Oberon for Windows v2.5, downloaded from ftp://ftp.inf.ethz.ch/pub/ETHOberon/Win95NT/PlugIn/ : WinOberon.zip + WinOberonFix.zip Unfortunately, it randomly crashes on startup (more often than not). The included debugger shows that there is an "access violation" happening in (Win32.)Threads.CheckStacks. The same crash happens for v2.4 and for v2.5 without the *Fix.zip installed. It is likely that the problem stems from the "sp=0" value, possibly at the SYSTEM.GET(sp, p) call. If I'm correct, then the problem might be caused by Windows 7 permission issues in Kernel32.GetThreadContext call. That's just a wild guess, though, for I don't know how to check it. And it's strange that once out of 10 tries the exception does not happen and the system runs normally. When it does crash, I see the main window appearing, maximizing itself, fully painting, and then Windows asks me if I'd like to report the bug to the developers. I'm running Windows 7 Professional. The Debugger's "Action - Show Trap" output: Debugger started Oberon ready access violation Threads.CheckStacks PC = 5406 bottom = 40960000 cont = 0404D98H ЮЏдъ.ontext (I5Lъ.ontext) p = 0 sp = 0 t = 01607A40H I5Lъ.э°@ (I5Lъ.hread) Kernel.GC PC = 4377 excp = 0409438H I5Lъ.xcpFrm i = 0 m = 00H (I5Lъ.oduleDesc) p = 4219104 ptr = 0404E38H I5Lъ.xceptionRecord (I5Lъ.lockPtr) ptradr = 4200560 Oberon.GC PC = 6411 me = 01607CA0H I5Lъ.и% (I5Lъ.askDesc) Oberon.OLoop PC = 13214 M = 0457558H I5Lъ.nputMsg V = 016A1660H I5Lъ.nputMsg (I5Lъ.iewerDesc) lastTask = 01607CA0H I5Lъ.и% (I5Lъ.askDesc) time = 1084556749 Threads.Wrapper PC = 3780 excp = 04044D8H I5Lъ.xcpFrm id = 3892 par = 0 pt = 00H (I5Lъ.hread) t = 01607A40H I5Lъ.э°@ (I5Lъ.hread) wrapper = Threads.Wrapper PC = 3486 Unknown EIP = 077A21174H EBP = 0270FF94H Unknown EIP = 0778DB3F5H EBP = 0270FFD4H Unknown EIP = 0778DB3C8H EBP = 0270FFECH Is there a fix available? Is there anyone working feverishly on the solution? Alternatively, what can I run if I don't want to reboot or install virtual machines? Can Aos run as a Windows 7 app? ---=====--- Alexander

4 7

Re: [Oberon] A slightly better procedure
by Riza Dindir 13 Dec '10

13 Dec '10

Hi All, On Sat, 11 Dec 2010 23:28:31 +0100 Jan Verhoeven wrote, > Hi Nemo > On Saturday 11 December 2010 16:32:07 Nemo wrote: >> A (minor) visual improvement is to write ('9' >= ch) & (ch >='0'). >> Of course! Just like we do in mathematics! It reads a lot faster. >> This assists in comprehension as you see the upper and lower bounds >> of the interval at once. (I have known companies with this rule for >> programming languages and it works well.) > Thanks for the hint! IMHO it might be cleaner to write ('0' <= ch) & (ch <= '9'), by moving the lower bound to the left. Mit freundlichen gruessen, Kind Regards, rd

1 0

A slightly better procedure
by Jan Verhoeven 11 Dec '10

11 Dec '10

Chris hinted me to the following improvement. = = = = = 1. BOOLEAN expressions ---------------------- PROCEDURE IsDigit (chr : CHAR) : BOOLEAN; BEGIN IF chr < '0' THEN RETURN FALSE END; IF chr > '9' THEN RETURN FALSE END; RETURN TRUE END IsDigit; It is not a good idea to get into the habit of having multiple exits from a procedure. In this case they are unnecessary as the recommended Pascal / Modula-2 / Oberon programming style would result in: PROCEDURE IsDigit(ch: CHAR): BOOLEAN; BEGIN RETURN (ch >= '0') & (ch <= '9') END IsDigit; = = = = = = which can be further improved to PROCEDURE IsDigit(ch: CHAR): BOOLEAN; BEGIN RETURN (ch <= '9') & (ch >= '0') END IsDigit; since there are many more characters above '9' than below '0'. -- Met vriendelijke groeten, Jan Verhoeven http://www.verhoeven272.nl

2 2

A first experience with obc
by Jan Verhoeven 08 Dec '10

08 Dec '10

Hi friends, Here http://fruttenboel.verhoeven272.nl/obc/triples.html is a first experience with the obc compiler. It may also be used by people getting to know Oberon (like I am doing right now). -- Met vriendelijke groeten, Jan Verhoeven http://www.verhoeven272.nl

3 3

How to mimic an associative array
by Duke Normandin 06 Dec '10

06 Dec '10

Hello list... Given that Oberon-2 does not natively support associative arrays (aka hash; map - in other languages), I need to come up with a tailor-made equivalent. Still very much the Oberon/Modula/Pascal noob, I thinking that maybe a Record like: TYPE myHash = RECORD key : ARRAY <someLength> OF CHAR; value : ARRAY <someLength> OF CHAR; END; Would that work? Is there a better solution? I'm going to test the above, of course, but I just thought that I'd get your input, before spinning my wheels too much. There tires/tyres and head are already bald enough. :) -- Duke

14 63

Re: [Oberon] XDS Oberon and svgalib
by Frans-Pieter Vonck 02 Dec '10

02 Dec '10

Hello Alexander, > If it's a DLL, I use the Linux version of XDS. For windows there is a good graphical library available: Graph At the university of Brussels they use it at an introductory course in programming http://parallel.vub.ac.be/education/modula2/projectI/studentwork.html Any suggestions for XDS Linux and SVGAlib? Greets, F.P. > Hello! > > FPV> I'm trying out the XDS Oberon compiler. What I like to do is to use FPV> SVGAlib with XDS Oberon. Does anyone have tried importing C-libs into XDS > FPV> Oberon? > > If it's a DLL, then yes, sure! > http://uranus.ru/download.html > > ---=====--- > Alexander > > Frans-Pieter Vonck Jesse Owenshof 69 1034WT Amsterdam 020-5374854 06 40343011 Frans-Pieter Vonck Jesse Owenshof 69 1034WT Amsterdam 020-5374854 06 40343011

1 0

ETH Oberon command-line compiler
by Alexander Iljin 02 Dec '10

02 Dec '10

Hello! I'd like to announce that a command-line compiler project was recently published. Basically, it's the OP compiler from ETH Plugin Oberon for Windows Release 2.4 packaged as a command-line tool by Eugene Shcherbatyuk (I tried to contact him, but his e-mail address apparently now belongs to someone else). I got the package from Andras Pahi over 2.5 years ago, and it recently became interesting to me. I amended the package so that compiler can compile itself, fixed a couple of minor bugs and added/updated sample projects (Hello*). The build tool of choice is GNU Make, but simple bat-files are included as well if you don't have Make installed. If you do want to use Make you will also need the GNU tr utility for a successful build. Pre-built executables of the compiler and linker are included as well. I also added a linker option '/d' to output a list of project dependencies extracted from a given Link-file. This is used to automate building with Make. The project is hosted at GitHub, and the VCS of choice is Git. Project page: https://github.com/AlexIljin/OPCL Git link: git://github.com/AlexIljin/OPCL.git Git VCS: http://git-scm.com/ I have some plans on udpating the project. One of the tasks is to update the compiler to its current state from the latest version of Plugin Oberon I got from Emil Zeller recently (add MMX/SSE/SSE2 and delegates). Suggestions and contributions are welcome, please, use this simple issue tracker: https://github.com/AlexIljin/OPCL/issues Also, if you have any questions about Git, the project, etc, please, don't hesitate to ask. ---=====--- Alexander

2 2

Re: [Oberon] How to mimic an associative array
by Duke Normandin 02 Dec '10

02 Dec '10

On Thu, 2 Dec 2010, Chris Burrows wrote: > >-----Original Message----- > >From: Duke Normandin [mailto:dukeofperl@ml1.net] > >Sent: Thursday, 2 December 2010 10:18 AM > >To: Chris Burrows; ETH Oberon and related systems > >Subject: Re: [Oberon] How to mimic an associative array > > > >> "Algorithms and Data Structures" (Oberon Version Aug 2004) > >by Niklaus Wirth. > >> You can download a copy (1.2 MB PDF file. 179 pages) from: > >> > >> http://www-old.oberon.ethz.ch/WirthPubl/AD.pdf > > > >I have a copy, and I looked at it this morning. Scary! It's > >the presentation, NOT the subject matter. If only I wasn't so > >mathemata-phobic. :D > > > > I just had a closer look myself. I agree that Section 5.4 is scary but I'm > happy just to take his word for it: > > "What we wish to be assured of is that on the average the number of probes > is small. The following probabilistic argument reveals that it is even very > small." > > (Translation: "Here's proof that it works very efficiently.") > > and wait for somebody else to dispute his reasoning ;-) > > As for the first three sections, they are very precise and concise but not > nearly as scary. Reading them is a bit like eating rich food - if you gobble > it down you'll get indigestion. Consequently, here's a couple of suggestions > that hopefully that will make it easier to absorb: > > 1. Print out the pages - don't attempt to read them off a screen. > > 2. Spend as much time reading each sentence as you would normally take to > read a paragraph in a novel or a newspaper. > > Once you have done that, let us know if any particular sentence does not > make any sense. All good advise! I wonder though, if we're talking about the same thing? Somehow implementing a Perl-like associative array, using basic Oberon-2 data-types? As in: http://en.wikipedia.org/wiki/Hash_table -- Duke

2 2

How to mimic an associative array (2)
by Les May 02 Dec '10

02 Dec '10

Oberon people emphasise the virtues of simplicity. The original suggestion by Duke is easy to understand and simple to implement. Present day computers have huge memories so the size of the array to store these records is not an issue in most cases. Present day computers run very fast so search times are not likely to be an issue even if the array is quite large and a brute force method of searching for the key is used. Why use more complicated methods if a simple method will do the job? Les May

1 0

associative list
by jan＠verhoeven272.nl 02 Dec '10

02 Dec '10

Duke Is this http://www.pageresource.com/jscript/jarray2.htm what you're looking for? If so, I guess Niklaus Wirth would turn over in his grave (if he were dead by now, which he luckily isn't). It's against all the basic ideas behind Oberon. IF you would want to have such a feature, consider a dual stage method with an array to link a name to an index into a second array. You are right about the math involved with such a programmatic monstrum: only scripting tools like tcl offer this kind of inefficiency as a feature. Perhaps you get a clue from either one of http://fruttenboel.verhoeven272.nl/ack/dis06.html.gz http://fruttenboel.verhoeven272.nl/AVR/avrdis.html.gz http://fruttenboel.verhoeven272.nl/m4m/Plov21.html.gz The AVR processor has lots of GP registers with fancy names and each single bit inside that register has fancy names as well. So I use two linked lists for linking bit names to register numbers. -- Met vriendelijke groeten, Jan Verhoeven

1 0