I find myself, oddly in general agreement. See comments in context.
"Denver Braughler"

> > John Bertoglio wrote:
> To me, major releases are those that require actions like recompiling all
object code,
> reformatting the database, running some sort of conversion, or, code
changes.

Very important point. A true maintainence release would simply fix things
without major efforts. I might give a pass to recompiles of classes since
even the most trivial change usually requires this.

>
> Minor releases add new features but should not break mature code.
> A minor release (say, x.3) could supersede the preceding minor release
(x.2) where an
> error is being corrected that affects recent code using new features in a
manner that
> requires some re-coding, but does not affect mature code.
> In such a case, x.2 would be listed a "recalled".

Again, right on. If x.2 is really broken then it is reasonable, if somewhat
annoying, to have to take certain steps like recompiling your code. However,
as above, things like conversion should never happen at this level.
>
> Maintenance releases fix bugs or make small tweaks like another parameter
to existing
> features that are so isolated that the chance of adverse consequences to
existing
> installations is negligible.
> Probably they should be released as a patch to an existing binary
executable or as an
> XML export file.
>
This is how most people do it.

> 5.0.11 sounds too plump to be a mere maintenance release.

Good plumpness, but I agree. The real test is this: If I upgrade and nothing
(bad) happens, it is maintenance release.

>
> Exactly what 5.1 will require, I have not heard.
> But I also think that it probably warrants being classified as a major
release because
> if nothing else, existing procedures regarding user security may have to
be changed.

If the security is a superset of the existing system (that is, you can do
new stuff in new ways but you can continue to work as before), I would not
consider it a big change. If after I install 5.1 no one can get into the
system, that is a pretty big deal. For example, dropping support for Win
95/98 is not a big deal if the issue is just internal support. If it NO
LONGER RUNS of these systems, this is a big change and should be noted by a
signficiatn revision.

jb



Reply via email to