ietf-822
[Top] [All Lists]

RE: Enabled mail

1993-11-24 00:10:34
As I said before, you really need to take a look at the code before determining
whether or not your requirements can be met.

I will be, but using it still goes against "Rhys's Rule of Programming #256":

        Don't use someone else's code, unless the time necessary to
        write it yourself is going to be non-trivial.  You'll have the
        heartache of squeezing it into your user interface paradigm, have
        the hassle of keeping up to date with their new releases, and have
        the hassle of mutually incompatible copyright/license schemes.
        You have better things to do with your time.

Now, from what you say, TCL looks _very good_ with a minimal amount of
heartache, but that doesn't change the fact that I don't like using other
people's code.  Look at it this way: I'll be a test case to see if TCL
really is implementable from scratch within a reasonable (i.e. small)
amount of time.  If it isn't, then the MIME community may end up standardised
on a single-sourced TCL implementation.  An interpreter that can be whipped
up over the weekend by anyone competent in parsing and general interpreter
techniques (as I am) is much better than "just get the standard implementation
and port to your environment" IMHO.

Many people, especially in the Unix community, have this "don't reinvent
the wheel" philosophy.  Nothing wrong with that, but I prefer to have a
wheel that I control rather than trying to patch the bald spots on other
people's wheels.  That's just the way I am.  I am a little sour on Unix
programmers, being a veteran of integrating the free UnZip code into
one of my applications as a library: not an experience I care to repeat.
TCL may not be the same experience, but I don't want to be tied down to
someone else's implementation, no matter how good it may be.

Some of you (probably Ned) may say that I'll be forever playing catch-up
on TCL's features if I do my own.  Well, if TCL changes too much then
MIME shouldn't be using it in the first place because safety will be
impossible to guarantee. :-)

Really, I'm arguing from a position of weakness here because I have nothing
near the experience of Nathaniel and Marshall in active mail, but I'm
definitely dead against something that will take a non-trival amount of
time for average CS grads to write themselves.  The more non-trival things
we have in MIME, the more "exclusive" the club of people that have fully
implemented a usuable MIME system will become.  Now, there are some non-trival
things in MIME already: GIF, audio/basic, etc, but by their very nature they
are intended to be farmed off to other applications.  Not so with Safe-TCL,
which will work best integrated with the mail system itself.

I'll do my coding experiments and let you know.  Initial experiments tend
to point to the TCL core being fairly trivial: just the library routines
will take some time.  Maybe I'll be able to extinguish this storm in my
teacup.  Humble pie MIME attachments can be e-mailed to this address. :-)

Cheers,

Rhys.

<Prev in Thread] Current Thread [Next in Thread>