For testing pdfjam’s pdfpages integration, I ran into a few inconsistencies regarding whitespace in output such as
[2
<./source-1.pdf> <./source-1.pdf> <./source-1.pdf> <./source-1.pdf> <./source-1.pdf> <./source-1.pdf>] (a.aux)
Whether or not there is a space between the file names as well as whether there are line breaks varies depending on the tex engine and its version. Additionally the ids varied in output such as
<source-1.pdf, id=7, 597.508pt x 845.047pt>
These variations were quite nasty for ci/cd workflows running on a GitHub machine with an older texlive version than mine. By now, I fixed this by adding some own normalization machinery (just remove all whitespace after [1 until the ]) but thought that maybe I should bring this to your attention and ask whether you are interested in integrating any of this sort in l3build directly.
For reference: my normalization via `awk` and `sed`
<"$dir/a.log" awk 'BEGIN{s=0}
/^ *\[[0-9]/{ORS=""}
/]$/{ORS="\n"}
/] *\(/&&ORS=""{s=1}
/\)/&&s=1{s=0;ORS="\n"}
{print}
' | sed '/^ *\[[0-9]\+/s/ //g;/\.pdf,/s/id=[0-9]*/id=00/' >"$job.log"
If you are interested, I can rewrite this logic into Lua and create a PR.
For testing pdfjam’s pdfpages integration, I ran into a few inconsistencies regarding whitespace in output such as
Whether or not there is a space between the file names as well as whether there are line breaks varies depending on the tex engine and its version. Additionally the
ids varied in output such asThese variations were quite nasty for ci/cd workflows running on a GitHub machine with an older texlive version than mine. By now, I fixed this by adding some own normalization machinery (just remove all whitespace after
[1until the]) but thought that maybe I should bring this to your attention and ask whether you are interested in integrating any of this sort in l3build directly.For reference: my normalization via `awk` and `sed`
If you are interested, I can rewrite this logic into Lua and create a PR.