Re: Functions calls got slower between Octave 2.9.5 and 2.9.6 through .10

[John W. Eaton]

On 27-Apr-2007, Luis F. Ortiz wrote:

| On Fri, 2007-04-27 at 13:57 +0200, David Bateman wrote:
| 
| > Luis F. Ortiz wrote:
| > > On Thu, 2007-04-26 at 17:13 -0400, John W. Eaton wrote:
| > >> On 22-Apr-2007, Luis F. Ortiz wrote:
| > >>
| > >> | As you can see, something went awry between Octave 2.9.5 and 2.9.6,  
| > >> | and has stayed bad since, to
| > >> | the tune of 20% less goodness.
| > >>
| > >> The following patch will help slightly.  
| > >>     
| > > Sorry to be a pest this early in the morning, but I did not see the
| > > patch in a following message or this message.
| > > Did I miss it?
| > 
| > It's in the CVS,  I did
| > 
| > cvs diff -u -r 1.74 -r 1.75 src/ov-usr-fcn.cc > patch
| > 
| > and attached the result to this message
| 
| 
| Thanks for the tip;  I had updated by CVS checkout, but since I saw no
| corresponding 
| comment in Changelog, I was not at all sure.

The entry

  2007-04-26  John W. Eaton  <address@hidden>

        * ov-usr-fcn.cc (octave_user_function::do_multi_index_op):
        Only deal with varargout if ret_list->takes_varargs () is true.

should be in src/ChangeLog.

| I'll put it through the paces ASAP.

It will only make a minor improvement.  On my system, 2.1.73 (I didn't
go back and build 2.9.5, but I still have 2.1.73 installed, so I tried
that) ran your example in around 3.65s. Current CVS is 4.25s.  With
the patch it is 4.1s.

Also, removing the other two lines related to saving the call stack
reduced the time to 4.0s, but we need those so they can't be removed.
I don't know what accounts for the remaining difference.

jwe