-frame create- possible fail?

[skolenik]

I have a do-file that I run on two different machines. It consistently breaks in the same spot on one of them, while executing correctly on the other. The spot where it breaks is -frame create-, but the error message I see in the log file is "unexpected end of do-file" after the last command before that. I am suspecting that Stata fails to create that frame, but returns an incorrect or indecipherable internal error message, and then the do-file/logging mechanics does not find any better way of reporting that other than the above message.

I need to -set maxvar 32000- as I am working with fairly wide datasets; however they only consume a total of 1Gb, as I can check on the machine where stuff runs without errors, with the same data set. I have 16Gb on the machine where execution fails (and Stata's -about- says 12Gb available, so I don't see how getting even another 1Gb would be an issue).

Is there any good way to debug this?

[skolenik]

Maybe Alan Riley (StataCorp) has an informed answer...

[William Lisowski]

Is there any good way to debug this?

Are the two systems running the same operating system and the same version of Stata?

Is the do-file exactly identical on both systems, or is something edited, such as a directory path?

Have you tried to create a minimal complete and verifiable example?

[Sergiy Radyakin]

Stas, consider other possibilities too. I am getting this same message when the file contains mismatched quotes and other similar errors.
What was the previous command and what is the code fragment that fails?
Can you isolate the issue in a minimal reproducible example?

[skolenik]

The files running on the two machines are identical. They are both Windows, they both run slightly dated Stata 16.1 (31 Mar 2020). I checked for mismatched quotes and braces; since the file completes on one machine, they are all OK. I checked for special symbols that may have thrown Stata do-file parser off; all the EOLs are CR/LF as they are supposed to be. The data are restricted on both machines; I don't have full access to the second machine, so it is not a good use of the time of the person who has access to it to try and reproduce stuff. I agree that a reprex would have been ideal, but there is no good way to create it.