Upgrade to latest... always! - was "Why?"

[brian at aljex.com]

If fp would desist in including new features and altered behaviour in
patch-level updates you'd have a point. (the z in x.y.z)

Also we need to be able to trust that no major re-writes of code happen in patch
level updates even if the end result is supposed to produce the same outward
behaviour.

That's not the case, so it's a completely valid policy for people to refrain
from shipping or updating clients to the latest version as soon as it becomes
available.

Add new features and modify behaviour in minor or major version numbers only. In
patch-level updates only correct bugs, and only modify behaviour or re-write
code as necessary to fix bugs and _no_ new features. Even if a new feature
might be "free" as a consequence of some bug fix, still don't include it.

If we could trust that that was being done, only then could anyone say there was
no excuse for not staying up to date. Actually, it's *still* valid for people
to not take anyone elses word about breakage. And It's *still* true that
updating some windows sites is a costly process that no one wants to pay for or
do for free.


And while were talking about updates...

As long as the major version doesn't change, then bug fixes should continue
to be applied to all the minor versions. IE: if the current major version is 5,
and there exists 5.0.13 and 5.1.4 and 5.2.7, then a newly discovered and fixed
bug should be applied to to all extant minor versions not just the latest. That
is a new bug fix would cause the creation of 5.0.14, 5.1.5 and 5.2.8.

And the last version of all releases should be available either by download or
by cd (and it's perfectly OK to charge for shipping and handling for the CD and
it's perfectly OK if the charge needs to be more than $10 to really account for
someones time pulling out some old version and burning a cd and mailing it out)
IE: if I have a valid licence number for 4.5, then I should be able to get a
copy of the last update that was made to that version. This buisiness about
updates being created, but you being left out in the cold if you didn't happen
to download a copy during the window of opportunity, or you didn't happen to
download copies all your clients various licence numbers to get all the
different user counts and devel/runtime combinations... That basically sucks
and fp has no excuse for inflicting it on us.

OR, Really, if fp really wants to cast off old versions that completely and
pretend they don't exist then it must be ok to copy them once they are no
longer recognized as part of the fp universe right? I don't think fp should get
to have it both ways. Hmm, maybe the one valid obstruction would be that the
binaries include embedded strings that reference fp and fp no longer wants to
be responsible for them. So if we strip out the strings? Or modify them to
accredit fp for having done the original work but no longer supports this
version?

Of course it's not ok to copy 4.5 and I'm not actually advocating it. What I'm
saying is that I don't think it's right for fp to apply a double standard.

Any company, and *especially* fp in these lean times and competion-filled
waters, should be as user friendly as possible.



----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.