LG>> Many SCO OpenServer systems operate in vertical market turnkey 
 LG>> applications and the customer absolutely does not need that
 LG>> compiler present -- in fact, in many installations the presence
 LG>> of the compiler could be considered a potential security risk.
 MB>> That would be a strange variation on the theme "Security by
 MB>> Obscurity", which has already proven a bad solution many times.

 LG> Au contrare my friend

Btw, it's "au contraire".

 LG> Leaving the compiler completely off the system is more appropriately 
 LG> labeled "Security by Omission". :-)

Ok, I agree.

 LG> I can't comment on EMacs, Netscape, etc; but the X Window System itself 
 LG> has fundamentally held true to the form.

X is not a "small" tool. If it were, it wouldn't be 600k or 2200k.

 LG> I believe the development product is much better if the devsystem team is 
 LG> not constrained by the deployment schedule of the OS team. Wouldn't you?

I'd say any product is better if not constrained by any schedules.
"when is it going to be released?" "when it's ready."

 LG>> No..... there are other reasons, Mathieu.
 LG>> Please look at other perspectives other than the programmer's to
 LG>> see this point.

Many non-programmers do use a compiler often, on *nix systems. More
than ever. Should I pretend that they don't exist, or should I pretend
that they can read/write C code ?

 LG> Look at the ratio of
 LG> -application- systems to -development- systems and I believe you'll find
 LG> somewhere around 3/4 of all Unix installations are not of the type that 
 LG> would find a use for a development system.

What's the difference between an application system and a development
system? How is a spreadsheet language or a database language lesser
programming than C ? What makes a spreadsheet or a database NOT a
development environment? What makes writing SQL different from Real
Programming? Aren't computers general-purpose machines, and isn't a
compiler just another program, with ordinary requirements? (hint:
X plus Netscape takes more RAM than Vi plus C++ compiler.)

 LG> The "testing" cannot have been done if I'm the one
compiling the code.
 LG> TESTING cannot be done until AFTER the code is compiled.
 LG> If the product is "Tested", then that means I have an
inviolable binary
 LG> distriubtion and thus do not need a compiler.

Testing an application is about fixing the source code, not about
fixing binaries. Binaries get fixed by recompiling fixed source code,
and in many environments, compilation doesn't exist or is completely
hidden: Lisp, SmallTalk, Perl, SQL, Bourne shells / makefiles,
spreadsheets, etc. Once the testing is done, I hand out the fixed
source code, that has clear semantics relatively to all valid
implementations of the compilers and/or interpreters. Hope your
system's implementations of such, validly process the language.

 MB>> Well, any Unix system has a whole f_cking bunch of tools you will
 MB>> not need. If all you have is a static web server, will you remove
 MB>> "sort", "grep", "find", and all
those applications you don't need?
 LG> Perhaps. If I'm trying to idiot proof a machine that I'm the contract
 LG> administrator for and I'm not fully trustworthy of the inclinations of my
 LG> client's neo-tech-nerds to go screwing with a production machine.

If they needn't an account, don't give them one. Anyway, the machine is
probably networked in some way (Nfs,Ftp) that allows them to shoot
executables they need across machines, and this is not a great barrier.

 LG> If the archive management staff has "compiled the applications for all
 LG> supported platforms" then I'm not interested in the source
code - give me 
 LG> the compiled binaries.

This is a good point.

 LG> The reason that one has to download a source-code archive is one of two
 LG> reasons:
 LG> (1) There is no binary distribution available, or
 LG> (2) The downloader wishes to have the capability to tweak (modify) the
 LG> source code before recompiling.

There's at least the third reason of knowing the exact semantics just
by reading the source. This is important if you don't trust anyone.

 LG> Nice dodge -- but the question is valid. It establishes your
 LG> familiarity (or lack thereof) of how production application
 LG> systems are implemented in the majority of corporate installations.

Why is the Majority of any importance here? How does that prove the
validity of the methodology of their system builders? Why is this
Majority any more wise than the one waiting until late 1998 for
checking the systems for Y2K bugs?

 MB>> I'm not trying to be rude or to scare you off. I'm just not a PR
 MB>> person... and I don't think I express my ideas very well...
 LG> We both suffer that deficiency. :-)

wow, now that's cool! :-)

matju

--- Terminate 4.00/Pro