Re: Free Pascal 2.4.2 released!
In comp.lang.pascal.misc message <firstname.lastname@example.org
l>, Wed, 17 Nov 2010 11:46:38, Marco van de Voort
>On 2010-11-16, Dr J R Stockton <email@example.com> wrote:
>(discussion about webdesign skipped)
>> It would be good to know, at an early stage, what side-effects
>> installation may have.
>Yes, and if you know unknown interesting side-effects, do tell us.
>Besides the obvious changes to the global structure (add directories to path
>which might produce conflicting binary names in any scenario), and the file
>registrations on Windows, both of which can be disabled, I know only of one
Lengthening the Path can be a pain; I have utilities written in TP which
can only handle 255 character paths, and it appears that COMMAND.COM and
CMD.EXE have effective limits; the former seems to be 127. It should be
kept to a minimum. I have a theory that some applications, including
languages, are designed in the hope that they are the only thing ever
added to a new Windows system.
>on the win32 platform the textmode IDE includes cygwin gdb, a DLL that does
>a version check using a shared memory space. When the IDE is up, a cygwin
>installation using a different version of cygwin might complain and vice
>> ISTR that when I installed one free version of Pascal, it broke the way in
>> which I operated something that I liked to use - I no longer recall how.
>The typical risks of adding to the path are the only ones I can imagine, and
>they go for TP and Delphi too. Specially the make.exe binary is a problem
>if you install any development tool. Even Borland has some old own bastard
>substitute for that, in which I'm not interested, and it always obscures my
>more fully featured GNU make.
The TP installer may add to the Path, but TP5 does not need any such
addition if hand-copied.
>> And to know the disc space needed
>Depends on various installation choices, and exact target platform. The
>win32 install might need well over 100MB with all options installed. The
>Dos install can be hand picked and can maybe fit in say 10MB if you extract
>the zips by hand.
Even rough limits can be useful, as can the size of the download.
>> (I still own a machine with 2 * 720 kB
>> floppies, only; it runs Turbo Pascal 5 nicely, but TP6 is rather slow).
>Since it probably is not a 32-bit or 64-bit machine, it is excluded by
>definition, since for PC, FPC targets the i386 or x86_64 architecture. (and
>indeed it describes itself as a 32/64-bit compiler)
1988 Amstrad PPC640, OS Amstrad DOS 3.3, 1987. I just tested it, Turbo
Pascal 5 can at least calculate 6*7 on it. I'd not expected that
anything modern would run on it; merely quoted it as an indication that
there are still old machines around.
>> A thought : A Web page can upload a text to a server. If that
>> text were a Pascal program (perhaps with a length limit), might
>> it be possible for the server to compile that program and return
>> the results as either an error report or as something that could
>> be saved for execution?
>Yes. On universities that use Free Pascal for entry level CS courses, it is
>often even common practice that students upload their assignments. These
>assignments are then put through a series of tests and heuristics, and a
>database of students programs and automated test results are then presented
>to the lecturer or his assistants for grading.
>On the open internet this is more difficult though, since one could easily
>overload a server using such services. And it would have to be thoroughly
>sandboxed to avoid that running programs made from others can do damage.
I was not suggesting that the service would run the code, merely that it
would compile and return an EXE or error messages. I have neither the
knowledge nor the resources to make and run such.
(c) John Stockton, nr London UK. ?@merlyn.demon.co.uk BP7, Delphi 3 & 2006.
<http://www.merlyn.demon.co.uk/> TP/BP/Delphi/&c., FAQqy topics & links;
NOT <http://support.codegear.com/newsgroups/>: news:borland.* Guidelines