sagetex representation should align with sagemath CLI output, hence catch stdout

[meisterluk]

Expected output

sage: print('Hello World')
Hello World

Actual output

sage: print('Hello World')
None

Proposed solution

It is annoying that the output in sagemath CLIs/cells is different from the sagemath generated output.

• As far as I understand, eval and preparse are called to generate the output. The output is written to result. Then only result is represented, not what was written to stdout.
• With python3, print became a function and thus the problem occurs with one of the most fundamental primitives. Thus, it is IMHO particularly annoying.
• Simutaneously, it is easy to manipulate stdout, catch the written content and represent it in sagetex
• Thus, I propose to replace sys.stdout during execution, represent the caught lines and skip representation of the result if it is None

[kcrisman]

This is an interesting and potentially good idea (certainly the point about `print` is well-noted!). The main thing is to ensure that there is no subtle change in what people "usually" use SageTeX for. Are there any places where `result` is actually what is desired, not `stdout`, since this is intended for pdf/print and not interactive use? Or any concern that this might significantly increase processing time for a large project?

…

Message ID: ***@***.***>

[meisterluk]

I cannot speak for the entire community, but my personal take is …

• The output will change. Since there is no mechanism to communicate output changes across versions on CTAN, one might soften the blow with an optional flag. Per default, you get the previous behavior. Enabling the option, the output is caught as I described initally. If the community prefers it in general, the behavior will become default some years later and an option allows to get the old behavior.
• I submitted this bug report, because a user in a talk described how much annoyance this created for her. As someone with limited technical knowledge, she did not know what is going on. I looked into the technical details. So this comes from the community itself.
• I cannot see a problem with processing time. But the memory usage increases. This is an argument in favor of an optional flag instead of non-optional future behavior.

[meisterluk]

I learned that the request might be a direct result of Cryptool, a software and its German book about cryptography which uses sagetex for generating its content.