Great we are on the good way 
Could you test also something with brackets and another one with max function ?
I think only one pair of brackets is allowed and things get problematic when using more than one pairā¦it seems it is what i wrote in the small description of the process but just to sureā¦
Best
Not good, only very partial and not useful success:
(1.8*f46) success - but not as required
(f46*1.8) success - but not as required
Sorry, butā¦
(f46*9/5)+32 failed: āError: could not understand your formula SORRYā¦ā
32+(f46*9/5) failed: āError: could not understand your formula SORRYā¦ā
32+(f46*1.8) failed: āError: could not understand your formula SORRYā¦ā
32+(1.8*f46) failed: āError: could not understand your formula SORRYā¦ā
(1.8*f46)+32 failed: āError: could not understand your formula SORRYā¦ā
(f46*1.8)+32 failed: āError: could not understand your formula SORRYā¦ā
max(f46,f97) failed: āError: could not understand your formula SORRYā¦ā
OK normally for all those you donāt need the brackets, because multiplication and division take precedence over addition and subtraction. So 32+(f46*9/5) could be 32+9*f46/5 but I think now 32+f46*9/5 should succeed. And this pattern is quite intuitiveā¦but it is true in other ālanguagesā you can do this and the algo first removes the useless bracketsā¦donāt know if it can be achieved easily
brackets are useful for example when you do 32*(f32+10) because it is not the same as 32*f32+10
max was in my mind dealing only with feeds vs numbers, ie max(f31,0) will return only the positive values of the feed, or you can do max(f32-f12, 0)ā¦not sure it can be easy to allow max(f23,f45) I will check
True, but it is quite confusing for someone who does not know that unnecessary brackets are illegal, surely it should accept those brackets because what is inside has been proven accepted and correct. To fail ācould not understandā to me means no brackets, whatever is inside, are acceptable. This is why I tested no more.
I should have checked max(f46,0) first - this succeeds but I did not have the data to prove it, as does max(f46,5) which works correctly.
max(f97-f46,0) also succeeds.
If it cannot be achieved easily, we should document what is acceptable and what is not. The work-around is to do the calculation in parts to temporary (or virtual - Iāve only just thought of this?) feeds, then reprocess the temporary feeds to as many levels as necessary to get the final result - then throw away the temporary feeds.
@Robert.Wall : should be more robust now. Iāve made some corrections so that all the formulas you tested last time can be accepted and more.
Same process, test with the master branch of postprocess and weāll see
It is looking good ā all the tests in post no.22 passed this time. 