[Prev][Next][Index][Thread]
Re: Philosophy/strategy for CL-HTTP
Hi Robert,
In the case of the hyperarchive, we have a module that can be used
largely independently of the cl-http version. That's why
there are some patches to go with it for early adopters who want to
use it now. This is lisp. We can dynamically upgrade the system.
Others can wait until the next release.
With regard to the general philosophy, it is our intention to
maintain backward compatible functional interfaces. We have piles
of code that depends on it too. Thus, upgrades should merely involve
recompilation of applications.
With regard to the ideosynracies of platforms, we prefer using
configurations that work reliably and we will not recommend ones
that do not. The mac platform has become much more reliable for
networking lately and everyone is well-advised to upgrade. When
Digitool releases their native OpenTransport interface, PPC users
will be well-advised to upgrade.
As a research team, we're not going to devote our lives to supporting
legacy platforms. That is where commerical support comes in. In do
course, it will be available.
Whenever possible, we will leave the old code in place, but users who depend on it
will need to fix it when problems arise. A case in point is MCL 2.0.
We don't support it. If someone insists on running it, they can fix any
bugs too. MCL 3.0 is no longer in use by most developers, who by now
are using 3.1 and 4.0. If you encounter problems with 3.0, you could
report them along with code fixes to add to subsequent releases.
Finally, motion does not necessarily mean fragility, especially withb lisp. It means
fewer bugs and better performance. The Web standards are in a rapid state of motion.
We track (or sometimes) lead the standards. (Of course, motion in primitive operating
systems may well mean fragility.)
Dead projects with no development don't move. If you are afraid of change,
you can stick with some obsolete version. Nobody is making you upgrade.
I wouldn't worry to much about cl-http moving; think about what's going to
happen to the MacOS in the coming months. In another year, you won't need
to worry about any new bugs, unless you switch to the NextOS.
Ultimately, the support you receive is the support that you donate. Thus, users
will receive what they are willing to contribute now, and pay for later.
References: