Bug SS-192
1 vote

Calculation Restart at ComboBoxList select change

Created by ioan92 on 8/6/2014 10:18 PM Last Updated by Davide on 8/22/2014 11:47 AM
Logged: 0   (hrs)


With the new version of SMath 0.97.5329, it is confirmed an abnormal behavior:

* by changing the selection on a control as ComboBoxList, the calculation of the page is startet even if the automatic calculation is inhibited.

* this aspect is similar to that produced in the previous version when a new line was introduced.

This aspect is very disturbant in case of files with a long time of execution

Complementary informations are in this threds:







    Davide (Friday, August 22, 2014 11:47 AM) #

Hi Ioan :)


We have made a small "catch the bug" party, last night  with Andrey....  I am confident that a part of this bugs will be solved

- there is a reproducible problem when the mouse wheel does a long step instead of a small step

- another reproducible  problem on region disabled, when the region get and looses the focus (click on the region then click outside the region)

There are 3 bugs "on the fog", that I think that are related someway because are machine dependant... 

- the RequestEvaluation(true) with or without using doesn't work on my system but works on Andrey system...

- the mouse wheel action doesn't work on my system but works on Andrey system.

- the double evaluation of the ComboBoxList in your file (but not in others), that made me crazy; again as stated by Martin, works also for Andrey  :)

    ioan92 (Friday, August 22, 2014 4:11 AM) #

Hello Andrey, Hello Davide,


IMHO, we are marching fully in what Microsoft called some years ago as " DLL Hell".

I use older versions of Davide's plugins (see appened image) and with the newer SMath version (build 5346) there are no more problems of "compulsive" recalculation (!?).

Please check


Best Regards,



    smath (Thursday, August 21, 2014 11:39 PM) #

Davide, just checked new versions of your plug-ins and I see they works absolutely fine.

I didn't understood what you meant about "Enable" state, but regarding "doesn't trigger the evaluation until the region loose the focus" - it is by design. If you don't want to wait for focus lost you can request for immediate evaluation by calling RequestEvaluation(true).

Please let me know if it did help.

    Davide (Thursday, August 21, 2014 11:17 PM) #

Unfortunately I can't confirm. I've recompiled the plugins, the updated source are in the SVN, the updated libraries are in the EM.

I still experience the double evaluation on click and the not evaluation from the transaction (tested on the attached SS-191.sm).


- I've added a check to the "Enabled" state of the region before calling the evaluation: the output is not sent to the canvas (correct) but a recalculation is triggered O.o

- updating the RadioButtonList region seems that the timer doesn't trigger the evaluation until the region loose the focus (don't know if the issue affects also the previous version)

    smath (Thursday, August 21, 2014 4:08 PM) #


    ioan92 (Friday, August 15, 2014 12:17 PM) #

It seems that a recalculation is operated also after the introduction of a new calculation region.

    ioan92 (Thursday, August 7, 2014 10:30 AM) #

Thanks Davide,


Some complementary informations:

                     * both attached files are generating the restart calculation, even with a clean reinstallation, for me, in two machines Win 8 and 8.1;

                     * both controls (ComboBoxList and RadioButtons) are generating the phenomena;

                     * if I understood well, Martin confirms this aspects; he conssiders this as irrelevant because he uses SMath with automatic recalculation selected;

                        so he do not  noticed differences;

                     * the calculation seems restarting from the position of the control submitted to a change event.

    Davide (Thursday, August 7, 2014 2:08 AM) #

Is a really strange behavior. In the attachments SS-191.sm is a file made by Ioan. Here if you choose some profile from a comboboxlist (f.e. from the region  Profilés HE) I see an evaluation triggered twice. In the second attachment test.sm (a simplified case) this doesn't happens.

Moreover seems that alla works fine for Martin (same Framework, same Plugin, same SMath, both win7 x64)

As additional info I've experienced the issues in a clean installation (all files and registry keys removed with Geek uninstaller), and also in the clean portable version made from the linux archive.


Also, seems that the using( .... CreateTransaction() )  doesn't works anymore for me, but works for Martin... so really I don't know what could be the issue (damaged libraries in my framework?)