Re: Tcl/Tk 8.5a4 released

pedietz@west.com wrote:
>> * [TIP 215] Made [incr] auto-initialize.
 
> Isnt this also a POTENTIAL_INCOMPATIBILITY ??

Strictly speaking, every new feature (and even some bug fixes) carries
with it some possibility to be incompatible with some existing script.
It's a judgment call which of those get flagged with a special notice.
As we distill ChangeLog -> changes -> release notes, there's less and
less detail, and fewer such special notices.

No hard and fast rules, but we tend to think that things like
changing error messages, noting the potential name-conflict inherent
in a new command name, and changes that convert errors into successful
operations that properly generalize what was there before, are less
noteworthy than other changes.

On the flip side, we tend to put the warnings in place when we
know for a fact there exists code the change will break, often
because it's our own code examples that are breaking.

And simple error and insufficient review creeps into the mix as well.
For example, the 8.5a4 release notes fail to mention the new strict
errors when packages have [package ifneeded] commands with
versions that do not match their [package provide] commands,
though it's still there in the changes file:

2005-11-08 (bug fix)[1162286] [package require] checks that the script
registered by [package ifneeded] provides the version it claims (lavana,porter)
*** POTENTIAL INCOMPATIBILITY ***

We know there are broken packages out there that will hit this,
so it would have been better for me to give that one more prominent
mention in the release notes as well.

-- 
| Don Porter          Mathematical and Computational Sciences Division |
| donald.porter@nist.gov             Information Technology Laboratory |
| http://math.nist.gov/~DPorter/                                  NIST |
|______________________________________________________________________|