Gpc September 1997

gpc@gnu.de

28 participants
85 discussions

Re: Objects question...
by Peter Gerwinski 05 Sep '97

05 Sep '97

According to Frank Heckenbach: > > I think he meant dynamically loading and freeing the code of the object's > method, not it's data. AFAIK, one needs to use dynamic linking in order to > achieve this, but I don't know much about it. I didn't refer to data only: Modern DPMI managers swap both code and data to disk if there is not enough physical memory. On such a platform just forget about overlays and such: You always have them automatically. (This holds for Borland Pascal 7.0 in protected mode, too.) > BTW: Do DJGPP and/or EMX do this, too? Yes. Under DOS, DJGPP and EMX swap memory to disk unless you disable it. When EMX runs under OS/2, the OS does the swapping. Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal [970714] - http://home.pages.de/~gnu-pascal/ [970125]

1 0

Re: Objects question...
by Peter Gerwinski 05 Sep '97

05 Sep '97

According to Bernhard Tschirren: > > BTW - Code size is so minimal compared with DATA, I personally dont see ANY > reason for trying to dynamically load code. The OS (or DPMI server) does this automatically if needed and not disabled. > GPC executables are only so BIG because there is a HEAP of unused > library code appended to the end (BTW: WHY IS THIS??) See the DJGPP FAQ list, section 8.15. > A procedure is > usually not much bigger than 20KB!! That depends on how long your procedures are. Other than with BP, there is no limit in GPC how long a procedure can become besides your virtual RAM. In practise, I am often surprised hwo small my procedures get when compiled with optimization (-O, -O2, or -O3). Greetings, Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal [970714] - http://home.pages.de/~gnu-pascal/ [970125]

1 0

Re: Bug ??
by Frank Heckenbach 05 Sep '97

05 Sep '97

Stephane wrote: > Hi, my name is Stephane, I'm French and BP7 converted... to GNU GPC ! Hi! I'm also coming from BP7, like many of the GPC developers... > I think I found a BUG or something like that... > (file linked in ZIP) The problem is that your string constant was too long, and GPC didn't catch it. So the string extended to the memory area where the array was stored. This was fixed in a very recent patch which might not be publicly available. Just wait for the next release, and the bug should be gone! This means, you will get a proper error message then. :-) > {Another question :} > FOR I:=1 TO 400 DO S[I]:='X';{don't work why ?? (= Fillchar X)} What error shows up? It works fine with my version (provided, of course, that I declare the variable I). In your originial program, of course, you have the length of the string set to more than 400, so you'll get some "garbage" after the 400 X's if you write the string. Which GPC version are you using? If it's an older version, this might be an old bug which is fixed by now. The latest versions can be found at ftp://agnes.dida.physik.uni-essen.de/gnu-pascal/ Anyway, thanks for your bug reports, such reports help to improve GPC. Frank -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm

1 0

Bug ??
by Stéphane 05 Sep '97

05 Sep '97

Hi, my name is Stephane, I'm French and BP7 converted... to GNU GPC ! I think I found a BUG or something like that... (file linked in ZIP) Bye, The Kilobug Team

1 0

gpi mechanism
by Tom Dye 05 Sep '97

05 Sep '97

Hello, I just ran into the bug in the gpi mechanism that is documented in the gpc info file. My version of gpc is several months old. I have two questions. 1) Is the gpi mechanism fixed in the latest beta release? 2) If not, are bug reports on this still welcome? Thanks, Tom Thomas S. Dye, Ph.D. Home: 813 16th Avenue, Honolulu, Hawaii 96816. Voice (808) 734-2087. Work: International Archaeological Research Institute, Inc., 949 McCully St., Suite 5, Honolulu, Hawaii 96826. Voice (808) 946-2548; Fax 943-0716.

1 0

Re: Objects question...
by Orlando Llanes 04 Sep '97

04 Sep '97

From: Bernhard Tschirren <bernie(a)icenet.com.au> >BTW - Code size is so minimal compared with DATA, I personally dont see ANY >... >usually not much bigger than 20KB!! That's what I'm trying to avoid, there's a lot of programs that take memory for granted, and also, I don't want anything in memory unless it's needed. If that's the way it is, then that's the way it'll have to be I guess, I'm just trying to reduce unnecessary overhead. See ya! Orlando Llanes

1 0

Re: Objects question...
by Bernhard Tschirren 04 Sep '97

04 Sep '97

>> You are obviously writing a memory manager to keep the amount of RAM >> your program uses small and to use the disk instead. On UNIX-like >> platforms you don't need to do this because the operating system does >> it for you. > >BTW: Do DJGPP and/or EMX do this, too? > If you are using CWSDPMI as your DPMI host, DJGPP does it by default (up to 128MB or 256MB of virtual ram) - however it can be disabled (For games and such). PMODETSR (another DPMI host) does NOT provide virtual memory. If you run your program from Windows it will obviously provide swap space. I dont know about EMX but I would presume that it does the same thing. ________________________ e-mail: bernie(a)icenet.com.au | _) | | ptschirrenb(a)cc.curtin.edu.au |___)ernhard |_|schirren tschirre(a)cs.curtin.edu.au www: www.geocities.com/CollegePark/Quad/8070/

1 0

Re: Objects question...
by Orlando Llanes 04 Sep '97

04 Sep '97

From: Peter Gerwinski <peter(a)agnes.dida.physik.uni-essen.de> >You are obviously writing a memory manager to keep the amount of RAM >... >Your application does not even notice this. Sounds like you mean a Virtual Memory Manager, what I'm actually working on are the screen objects, and the reason I brought it up is because BP loads the object code whether it's needed or not. So basically, GPC loads all object code at startup? Or does it load it when it's called with New? See ya! Orlando Llanes

2 1

Re: Objects question...
by Frank Heckenbach 04 Sep '97

04 Sep '97

Peter Gerwinski wrote: > > Is there a way to > > create an object so that it will be loaded only if allocated with "New"? In > > other words, is there a way to keep the code for the object on disk until > > it's needed, and removed form memory when not? > > Write a constructor that fetches all needed data from the file when > the object is `New'ed; if you want to keep the stuff in the file > current, write a destructor that writes everything back to disk > when the object is `Dispose'd. (* BTW, this has nothing to do with > the internals how objects work at run-time. ;*) I think he meant dynamically loading and freeing the code of the object's method, not it's data. AFAIK, one needs to use dynamic linking in order to achieve this, but I don't know much about it. > You are obviously writing a memory manager to keep the amount of RAM > your program uses small and to use the disk instead. On UNIX-like > platforms you don't need to do this because the operating system does > it for you. BTW: Do DJGPP and/or EMX do this, too? Frank -- Frank Heckenbach, Erlangen, Germany heckenb(a)mi.uni-erlangen.de http://home.pages.de/~fjf/links.htm

1 0

Re: Objects question...
by Peter Gerwinski 04 Sep '97

04 Sep '97

According to Orlando Llanes: > > I'm curious about something with objects as it will affect the game lib. > What I need to know is how objects work at run-time. Each instance of each object has an implicit "vmt" field which contains a pointer to the "virtual method table" where the addresses of all virtual methods are stored. GPC produces inline code to initialize the vmt immediately before a constructor is called. (BP does the assignment inside the constructor instead.) When a virtual method is called, the program dereferences the vmt field, and so on. > Is there a way to > create an object so that it will be loaded only if allocated with "New"? In > other words, is there a way to keep the code for the object on disk until > it's needed, and removed form memory when not? Write a constructor that fetches all needed data from the file when the object is `New'ed; if you want to keep the stuff in the file current, write a destructor that writes everything back to disk when the object is `Dispose'd. (* BTW, this has nothing to do with the internals how objects work at run-time. ;*) You are obviously writing a memory manager to keep the amount of RAM your program uses small and to use the disk instead. On UNIX-like platforms you don't need to do this because the operating system does it for you. Just allocate as many MegaBytes as you need; if this exceeds your physical RAM the operating system swaps out the unneeded parts automatically and restores them when they are needed again. Your application does not even notice this. Have fun, Peter Dipl.-Phys. Peter Gerwinski, Essen, Germany, free physicist and programmer peter.gerwinski(a)uni-essen.de - http://home.pages.de/~peter.gerwinski/ [970201] maintainer GNU Pascal [970714] - http://home.pages.de/~gnu-pascal/ [970125]

1 0

← Newer
1
2
3
4
5
6
7
8
9
Older →

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

Gpc September 1997