>>>>> "Rupert" == Rupert Swarbrick <rswarbrick at gmail.com> writes:
Rupert> Robert Dodier <robert.dodier at gmail.com> writes:
>> On 2013-11-16, Raymond Toy <toy.raymond at gmail.com> wrote:
>>> I'm all for removing as many conditionals as possible, but do you
>>> think that's wise? Or are we dropping support for gcl older than
>>> 2.6.10?
>>
>> Oh, I dunno. Mostly I was just thinking out loud. I've never been
>> happy about all the #+gcl stuff and it made me happy to think about
>> getting rid of it. But there's really no hurry.
Rupert> Maybe we can have an unofficial "deprecation date". Something like "when
Rupert> a version of foolisp has been out for N years, we assume that all
Rupert> occurrences of foolisp are at least that new" ?
Rupert> I'd guess N=2 or so is reasonable (even Debian stable moves that fast!)
I don't think we need any hard rule. If maintaining support for older
versions becomes so burdensome that no one wants to do it, then
support is dropped. :-) That could be 6 month, or 5 years. (While
looking at the catch/throw issue, I think I tried cmucl 19a, which
still worked pretty well with current maxima. 19a was release in 2004,
I think. I suspect that is many lisps from 2004 would still work fine.)
Ray