[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: branch_release-1-4 brought into 21st century

From: Thien-Thi Nguyen
Subject: Re: branch_release-1-4 brought into 21st century
Date: Mon, 11 Mar 2002 14:55:53 -0800

   From: Marius Vollmer <address@hidden>
   Date: 11 Mar 2002 17:38:10 +0100

   What was wrong with the 1.4 release process specifically?

well, a process that produces an artifact that doesn't compile and then
leaves that artifact to sit like a steaming turd for over a year w/o
redress, shows there is disconnect between the process and the artifact.
this disconnect can be characterized by the kinds of questions you would
ask for any process (aka program aka algorithm):

  is this a real algorithm or is it some heuristic?
  what are the inputs/outputs (data interfaces)?
  if you run the process on same inputs, does it produce same outputs?
  how does it handle error conditions?

  where can i read its "source code"?
  how are patches to the source code integrated?

so, to be specific and general at the same time, the 1.4 release process
did not include enough reflection to support the valuing (much less the
practicing) of self-correction / self-improvement, and shirked Quality.

what is Quality?  need we ask anyone to tell us these things?

   Are you sure they wont go to 1.6?  They would have to eventually, why
   not now?  Why would they adopt 1.4.1, but not 1.4?

i'm not sure what anyone would do, that's why i ask them.  what people
tell me is that it's more important to ask them why they do things
because even they don't know what they will do, and are still trying to
figure out why as well, so i ask them that.  resulting discussion often
reveals that (like me), people are lazy and want to experience the joy
of programming w/ only their own self as limits, and make their
decisions accordingly.

"they would have to eventually" doesn't hold when faced w/ the buffet of
excellent scheme programming systems out there filling niches from
pedagogical to production.  guile is excellent, too, and guile users to
date have endured capricious abuse of their trust out of some kind of
loyalty to GNU or whatever, in order to get at the excellence, but this
doesn't mean they aren't once burned twice shy.

wrt 1.4.1, probably people would adopt it for bug fixes, and because it
requires minimal changes to their programs, so that they can continue
doing fun new stuff instead of the API retrofit drudgery.  they are also
probably waiting for tools to make migration easier.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]