@JosephWright after the usual round of errors due to typos (stengine) and wrong settings (pdfext={""}, don't ask me why this was in the build.lua) pdf testing seems to work fine ...
@UlrikeFischer There's an update for today: I'm still working on PDF normalisation (or rather getting the binary data out). Release some time this afternoon :)
@JosephWright The /Alt and /ActualText values should be encoded either in pdfdocencoding or utf16be. With pdftex I managed a more or less sensible output with \StringEncodingConvert (from stringenc) and \pdfescapehex. But stringenc doesn't work with luatex and I can't find something suitable in the luatex manual.
@JosephWright context has something in l-unicode.lua. But it is in bits and pieces and lots of renames (local X= y) and so quite some work to get everything together.
@JosephWright when you mentioned Bruno I remembered l3str-convert. But it doesn't work ;-(. ` \str_set_convert:Nnnn \l_tmpa_tl {öäü} {utf8} {utf16be}` gives (with luatex) the error Invalid UTF-8 string: missing continuation byte (x3).
@JosephWright Rereading the docu, the correct syntax is \str_set_convert:Nnnn \l_tmpa_tl {öäü} {} {utf16be} and this seems to give something more sensible. Now I have to check how to get the hex from it.
@JosephWright ` \str_set_convert:Nnnn \l_tmpa_tl {öäü} {} {utf16be/hex}` looks perfect ;-). Bruno is so good ;-)
@UlrikeFischer I've updated l3build on CTAN, and probably will switch to some macro stuff for a few days unless anything specific comes up with PDF testing
