Python 1.6 alpha 1 released

Fredrik Lundh effbot at
Sun Apr 2 13:50:30 CEST 2000

Bjorn Pettersen wrote:
> I haven't seen anyone claim that examples in the manual are not
> documentation...

in earlier releases of the library reference, some examples
didn't even work.  you still think they're formally binding?

> It would be very nice if "someone" could do that, but based on the
> rethoric I've seen from you on this issue, and the absence of comment
> from Guido, I certainly don't feel confident that it would be accepted.

you don't know that until you've tried it...

you may have to work on your presentation, though. thinking
about the following may help you come up with a stronger case:

-- how does typical socket.connect code look?

-- what exception do you get after this change?  how
   does the error message look?  can anyone else figure
   out what's wrong?

-- is there any chance that people might be expecting
   the resulting exception from socket.connect (or the
   code invoking it)?  or in other words, is it likely that
   there will be a try/except clause around the call, that
   will misbehave (thus hiding the actual error message)?

-- can anyone but the original author fix the problem
   without breaking the program?

-- is socket.connect used in all kinds of scripts and
   applications, or mostly in support libraries?  is the
   number of affected programs equal to the number
   of code changes that have to be made?  (if your
   answer is "yes", please tell us why).

-- the user base is growing exponentially.  so is the
   number of python implementations.  given that,
   when is the right time to tighten up the core inter-
   faces?  now, later, or never?

when you've done this, replace "socket.connect" with
"list.append", and try again.

if you still think that this change is a disaster, please submit
a patch, and a rationale for accepting it.  all you have to do
is to look for NO_STRICT_LIST_APPEND in the 1.6a1 sources,
and copy that code to the socket module.

shouldn't take you more than 10 minutes, or so.

(btw, don't forget updating the documentation).


More information about the Python-list mailing list