Re: Tcl/Tk 8.5a4 released



pedietz@xxxxxxxx schrieb:
Yes, this is a potential incompatibility.

The CHANGES log should reflect this...
Yes.

But as a small side note, take a look at the 'incr' man page in your Tcl
8.4 installation (or see http://www.tcl.tk/man/tcl8.4/TclCmd/incr.htm ).
Nothing there mentions that 'incr' throws an error when the variable
does not exist. So basically your code depends on undocumented behaviour
and the rule is that undocumented behaviour is subject to change even in
minor releases.


while we're at it, lets have incr THROW an exception when a total is >
100.
For no reason other than to make people add catches() and remove
catches() after every TCL release.

Sounds frustrated. Maybe i sounded a bit too harsh.

The fact that some points get missed during discussion is not uncommon,
all the -nocomplain options are evidence of those 'missed points' during
API design. One could have added 'incr -nocomplain foobar 1' instead, yes.
If you feel or think that the TIP 215 implementation is breaking valid
assumptions about what should be allowed to change in Tcl point releases
and what should not change you are free to raise your concerns on the
TCLCORE mailing list. Tcl 8.5 is still in alpha stage an the API isn't
frozen yet, so _now_ would be the right time to complain if you care.
Even better, look at the currently 'draft' TIPs and see if something
there is so horribly broken that you absolutly do not want it in Tcl and
add your comments...

One other point in the Changes list which could be flagged with
potential incompatibility is 'glob -nocomplain', but thats more of a bug
fix.

Michael







.



Relevant Pages

  • Re: [incr] and counting occurrences
    ... Just allwoing a non-existing variable as argumet of incr would solve the ... before the increment is added. ... strings this is not even a new concept in Tcl. ...
    (comp.lang.tcl)
  • Re: Hecl (was: ANNOUNCE: The Jim Interpreter)
    ... > They are like upvar, but, IMO, clearer, because upvar uses the name to ... > incr &foo ... Actually, using incr that way in tcl, will make incr look ...
    (comp.lang.tcl)
  • Re: Tcl to C or C++
    ... Stefano Chiappa wrote: ... > launches Tcl interpreters and interacts with them. ... because someone wanted an incr that works also on doubles (or complex ... compiled extension, or (if you know that there is a C compiler in the ...
    (comp.lang.tcl)
  • Re: incr initialization (was: Hecl)
    ... >> Actually, using incr that way in tcl, will make incr look ... I would suggest instead changing tcl to require you ... This would 100% reliable plus catch your ... errors at compile time rather than run time. ...
    (comp.lang.tcl)