shellesc: luatex logs `runsystem(<cmd>)...executed` without ending period

[muzimuzhi]

Brief outline of the bug

When executing a shell command, each of Knuth TeX, pdfTeX, and XeTeX logs runsystem{<cmd>}...executed. with a trailing period after executed. But the emulation in shellesc for luatex doesn't log that period, see line 204. https://github.com/latex3/latex2e/blob/aa89b361a668a483e249839f97e3280dcc316879/required/tools/shellesc.dtx#L194-L211

Minimal example showing the bug

\RequirePackage{latexbug}       % <--should be always the first line (see CONTRIBUTING)!
\documentclass{article}
\usepackage{shellesc}

\begin{document}
\ShellEscape{pwd}
\end{document}

Log file (required) and possibly PDF file

shellesc-trailing-period.pdftex.log shellesc-trailing-period.luatex.log

[davidcarlisle]

@muzimuzhi so it does.

We can of course add . something like

$ diff shellesc.sty~ shellesc.sty
95c95
<        .. ")...(" .. msg .. ")\string\n")
---
>        .. ")...(" .. msg .. ").\string\n")
99c99
<       .. ")...executed\string\n")
---
>       .. ")...executed.\string\n")
103c103
<        .. ")...failed " .. (msg or "") .. "\string\n")
---
>        .. ")...failed. " .. (msg or "") .. "\string\n")

But you still get different messages in fail cases, eg without --shell-escape

luatex then gives

runsystem(pwd)...(Command execution disabled via shell_escape='p').

with a . but pdftex gives

runsystem(pwd)...disabled (restricted).

as I have the luatex message (in msg in the code above) as a Lua string I could try to do string replacements to normalise the messages, but the luatex one is more informative so I'm inclined to leave it.

The difference only matters in l3build or similar (where I see this report orginated) and there Command execution disabled via shell_escape='p' could be normalised to restricted in a custom normalisation step if you wanted to call failing cases and reduce engine log differences.

[muzimuzhi]

Thanks for your detailed comparison for all three cases. I saw the cases but didn't test them. Yes I think for the other two cases the log differences can be normalized by a user-redefined \ShellEscape@Lua or perhaps by l3build itself.

To be honest at first I thought inclusion of runsystem lines in log-based l3build testfiles was a mistake. But then I saw for tests need \showoutput, runsystem lines are not easily avoidable.

[davidcarlisle]

OK thanks for the reply. I'll (just) add dots and not do further normalisation here.