home *** CD-ROM | disk | FTP | other *** search
- Path: nntp.teleport.com!sschaem
- From: sschaem@teleport.com (Stephan Schaem)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: toy operating systems, like AmigaOS
- Date: 20 Apr 1996 16:21:30 GMT
- Organization: Teleport - Portland's Public Access (503) 220-1016
- Message-ID: <4lb2ua$dun@nadine.teleport.com>
- References: <skllsf.984983.4.5@groomlake.mil> <4kplgo$o9r@canyon.sr.hp.com> <4ksu6q$haa@nadine.teleport.com> <19960417.7BBC9E0.3B7F@asd01-01.dial.xs4all.nl> <4l6037$9la@nadine.teleport.com> <31774642.46FE@info.unicaen.fr>
- NNTP-Posting-Host: kelly.teleport.com
- X-Newsreader: TIN [version 1.2 PL2]
-
- Samuel Devulder (devulder@info.unicaen.fr) wrote:
- : Stephan Schaem wrote:
- : >
- : > Jeroen T. Vermeulen (jtv@xs4all.nl) wrote:
- : >
- : > : In article <4ksu6q$haa@nadine.teleport.com> sschaem@teleport.com (Stephan Schaem) writes:
- : > : >
- : > : > This tell that the amiga OS is bad for programmers in genreral.. more
- : > : > or less depanding on the methode you choose to write code.
- : >
- : > : It means that the Amiga OS is a _bad debugging tool_--much like my wristwatch is
- : > : a poor screwdriver. An important factor in user-friendly, desktop OS's is that
- : > : the user is not a programmer and the two may have different requirements.
-
- : > I dont code in C on the amiga, but I sure running all those hack like
- : > enforcer, mungwall, etc... help reduce crash. but I find that little,
- : > very little.
-
- : Well, may I suggest to check out APurify on aminet (a program of mine :-).
- : It is something like unix Purify (if you know it). By the way, a new
- : version of APurify will be released soon (I hope).
-
- I haven't used purify on unix system.
-
- : But I agree, Amiga-os is missing some security features for programmers.
- : Memory protection is one of those missing things, but I'd like to add
- : a dynamical stack management. That could be achieved by using a MMU to
- : track stack overflow and increase stack size when a MMU is available. When
- : there is no MMU, the scheduler could test the stack and do the expansion
- : (it should not cost a lot more cycles since it already has to fill exec
- : fields). The last method is not 100% safe, but it is certainly safer than
- : nothing.
-
- I really hate programs that dont know their stack requirment and just
- crash because they dont check the default size.
- This feature to me is nice for progy that have size can fluctuate alot
- from 'small' to 'big'... Basicly saves you some memory at times VS
- setting your stack to the 'big' size all the time.
-
- My programing methode on the amiga require VERY little stack space, 4k
- is enought even for complex projects.
-
- : Resource tracking is also a nice thing, because I find strange that when
- : breaking some programs with ^C, its windows is left on the screen. That
- : is dangerous since the name of the window is a pointer to a string in the
- : code an since the code is no more present in memory, it displays some
- : garbage. That could lead to potential guru later.
-
- ^C work because the programmer is waiting to receive it, and so is aware
- to terminate all of its related task and free all related resource.
- If this do not happen, the program as a bug.
-
- Something that would help to have remtask() work like a unix kill is to
- have task register a kill() function... with that kill function doing
- its own resource tracking. Its not perfect, but when the task crash 99%
- of its own resource could be freed. Its a bit like a 'fail' exit point
- function.(You can goto it at anytime to exit prematurly the function and
- cancel it)
-
- For those windows I think you can find hacks to close them.
-
- Stephan
-
-