Modelica.Utilities.Files.removeFile doesn't work

[casella]

Run the attached model, which opens a file, writes something to it, closes it and then removes it. At the end of the process the file is still there, yet no error is generated.

[sjoelund.se]

removeFile("file.txt"); is not part of the flattened code for some reason :) I assume because it was not marked impure in MSL.

[sjoelund.se]

Seems like it is marked impure in MSL, but not printed if the function has the impure annotation. Would need some more time to figure out what was wrong...

[sjoelund.se]

See also ​m:#1886. We now remove the print functions, but the other parts probably behave as expected.

[casella]

Can we implement the simple heuristics proposed ​here?

[sjoelund.se]

Nope. We don't know if the function calls external functions...

[casella]

Too bad. BTW, I understand the currently impelmented heuristics is that functions with no output are considered impure. However, I have some test cases where the behaviour changes for better if I also add the impure keyword to the declaration.

[sjoelund.se]

No, functions with no output are not considered impure. Because then you could not call print, hashtables functions, etc as used in MetaModelica.

[casella]

Sorry, I am totally confused then. What kind of heuristics is currently available in OMC to detect impure function calls with constant inputs and avoid optimizing them away at initialization?

[sjoelund.se]

Pretty much none. We look for __OpenModelica_Impure=true, or we run with -d=nogen which disables generation of dll's (but since some external functions are structurally necessary, some external calls are allowed and are handled by -d=evalfunc).

[casella]

Aha. On 22 Feb 2016 9:25 you wrote me on Skype

I think we detect that they don't return anything, so we consider them impure :)

I was misled by that...

[sjoelund.se]

Well. I think we try to not evaluate them (at least in some instances). But not by considering them the same as impure (since impure functions are more special).

[casella]

Which is even worse from the user's point of view, because sometimes they work as expected and sometimes they don't, without any obvious pattern...

[sjoelund.se]

Milestone pushed to 1.9.5

[sjoelund.se]

Milestone renamed

[sjoelund.se]

Ticket retargeted after milestone closed

[sjoelund.se]

Milestone moved to 1.12.0 due to 1.11.0 already being released.

[casella]

Milestone moved to 1.13.0 due to 1.12.0 already being released.

[casella]

Rescheduled to 1.14.0 after 1.13.0 releasee

[casella]

Releasing 1.14.0 which is stable and has many improvements w.r.t. 1.13.2. This issue is rescheduled to 1.16.0

[casella]

Retargeted to 1.17.0 after 1.16.0 release

[perost]

This works with the new frontend now which doesn't remove any of the calls and marks functions as impure correctly.