From: Mike Thomas" <miketh at brisbane>
> Last week, Mike, you asked me to follow on from my earlier reports and check
> the remote math server stuff with your revised XMaxima interface. I
> apologise for my late reply but I am caught up in some urgent work at
> present (and for a while to come) and have had little time to help out.
Thank-you for this - I can't possibly test what I'm doing enough
and rely on the great testing you did to stay on track.
> Unfortunately I have lost track of which version to use, so I cut to the
> finish and tried today's Maxima CVS version, built with today's CVS version
> of GCL. These tests were run under Windows 2000 on a PII 333 MHz CPU using
> the stable version of Active State TCL (8.4.3). Generally everthing went
> well.
>
> In particular I did not observe any of the socket problems previously
> reported. I wonder whether the version of TCL which comes with the
> executable XMaxima installation has troubles in that regard?
Actually I think you did, but a bug in the code (now fixed) stopped
you from seeing this; I'll explain below.
> Pretest Hacks:
>
> After installation (make install, with configuration prefix set to
> "g:/maxm"), I renamed the "bin/xmaxima" script to "bin/xmaxima.tcl", renamed
> "bin/maxima" to "bin/maxima.sh" and then inserted a version of Dave's
> minimal "bin/xmaxima.bat" as follows:
>
> ------------------------------------------------------------------------
> rem Trival maxima.bat
> set MAXIMA_PREFIX=..
> start %MAXIMA_PREFIX%/lib/maxima/5.9.0rc1/binary-gcl/maxima.exe -eval (run)
On all Windows files, we should get in the habit of adding
%1 %2 %3 %4 %5 %6 %7 %8 %9
to the end of the command line, in case what we are starting takes arguments.
> Tests:
>
> Then I double clicked on the renamed "xmaxima.tcl" which started the Xmaxima
> shell and a separate DOS shell with maxima running in it, whose stdin and
> stdout appeared to be redirected to the XMaxima shell (this being an
> artifact of the batch script I presume).
Could I ask everyone to view having to use a .bat file on Windows as a bug -
xmaxima.exe should work without it, and it should set things like
MAXIMA_PREFIX for the maxima to run "out of the box". If MAXIMA_PREFIX isn't
set right by xmaxima then it's xmaxima's (my :-) fault and a bug.
> The examples in the XMaxima primer seemed to work OK, as did the
> "Help/Maxima Help" menu
>
> When I tried the other Help menu options, however (eg "Help/Project Page") I
> got Tcl errors as follows:
>
> ---------------------------------------------------
> couldn't execute "start": no such file or directory
> couldn't execute "start": no such file or directory
My fault - a Win98 hack that doesn't work on Win2000 it seems.
I'll fix it but not a show stopper.
>
> Trying some of the netmath stuff at U. Texas in "wfs/courses" (specifically
> the ODE stuff) went perfectly eg
>
> http://www.ma.utexas.edu/users/wfs/courses/427k/427k-1.html
>
> and
>
> http://www.ma.utexas.edu/users/wfs/courses/427k/nthtofirst.html
>
>
> These pages are a tour-de-force of XMaxima's capabilities - I wish I had
> been able to do this when I studied DE's!!
>
>
> Other stuff eg:
>
> http://www.ma.utexas.edu/users/wfs/courses/408c/plotnewton.html
>
> resulted in messages such as:
>
> -------------------------------------------------------------
> can't read "mess": no such variable
> can't read "mess": no such variable
> while executing
> "M "$mess:\n%s\nYou may need to Restart" $err"
> (procedure "sendMaximaCall" line 25)
My fault, now fixed, but this code is for when you hit a socket failure.
Because you were talking to a remote server, it could be either a
run-of-the-mill socket failure, or a more serious problem in the code
with timeouts.
What I have raised in the list as "the socket bug" is when this
occurs with the local maxima, which is still done by sockets, but
socket failures should be extremely rare. For the local maxima, it usally
means that the local maxima has died, and if so, you get a slightly
different error message:
...You must Restart..."
> Maybe those pages are defunct, or the servers they are calling on?
To tell, you would have needed to see the rest of the error message
which a bug stopped you from seeing (You can fix the bug by changing the line
set mes "Error sending to Maxima:"
in sendMaximaCall to
set mess "Error sending to Maxima:"
)
> You probably don't need this, but for the record, another negative comment
> is that maybe the plot menu options need attention (if I I recall correctly
> this never worked right anyway). For example clicking on the "Save" menu
> option in plot windows caused a set of axes to be drawn over the plot, but
> nothing seems to be done. For example, I don't get a requestor asking for a
> file name to save and it took me a while to work out that the file was saved
> to a default file name.
I'm working a little in the dark a little here because I never used
xmaxima until just before I started working on it. I know maxima
well, and it could be that WFS's use of Tcl grew out if a coversation
I had with him in the early 90's. But I don't know what never worked
right anyway, and rely on testers such as yourself to let me know if I
break things. I'm still learning the code. I fixed one of these "silent
saves" in the Browser file menu, and I'll fix this one too.
> It would be good if there was more control over labels etc, but this is not
> a show stopper in any case.
Can you post this as a Request for Enhancement in the Bugs list of the
Sourceforge site? As we get to a working version, it would be nice
to have lower priority bugs posted to a more organized forum.
(Personally, I'd like to see this utexas mailing list split into 3
lists, perhaps on sourceforge: the canonical users/developers/announce.
I'm finding this one-big-page-that-isn't-searchable apporach
furstrating. Or could we bidirectional gateway it to our won
newsgroup, either under gnu.* or alt.*)
> Running the tests gave a bunch of good results until:
>
> "Could not find `trigrat' using paths in FILE_SEARCH_LISP,SYSTEM"
>
> This is probably a result of the bodgy way I started XMaxima.
>
> Restarting Maxima allowed me to continue unscathed.
>
>
> Summary:
>
> Lastly, I would just like to say thanks for the great work you are doing on
> XMaxima - it really shows off what Maxima can do and is well worth saving in
> my opinion.
Great; glad to see people like it. I really need the explicit testing
reports like you provided - it really helps. Could you fix that 'set mess'
bug a retest netmath a little? I'm curious about socket bugs, because I know
some of the code is defintely broken withr egard to socket timeouts.
> If that Netmath site is kept alive then it should be a real winner for
> Maxima (and for maths students around the world) - I had a lot of fun there.
I'll also try testing more against netmath - I've been foussing on local.
I'm getting a failure in rtest15.mac - WinME 5.9.0rc1 maxima code:
..Which was correct
/* ********************** Problem 4. *************** */
%Input is
1
INTEGRATE(SQRT(- 2 + - + x), x, 0, 1)
x
The result is
4
- -
3
Using ratsimp
Using ratsimp
This differed from the expected result:
4
-
3
Mike.