@samcarter secret ping :)

@PauloCereda secret reply :)

@samcarter awwww <3

@PauloCereda ooh, secrets

@Skillmon ooh

@PauloCereda secret breakfast

@DavidCarlisle oh no

@Richard Hi Richard! Tom from Overleaf Support here :) Nothing should be stopping you from doing that from what I remember as our environment is pretty permissive and you can do a lot there. But TBH we don't deal with this sort of things often, so I may remember things wrongly...

@JosephWright would I be expected to get different test logs if check is isolated? for example, baskervaldadf gives me

*** ../../build/test/baskervald-auto-test.tlg   2024-11-11 04:14:33.288428837 +0000
--- ../../build/test/baskervald-auto-test.pdftex.log    2024-11-11 04:14:43.016359873 +0000
***************
*** 3611,3616 ****
--- 3611,3617 ----
  ....\glue 0.0 plus 1.0fil
  ....\T1/ybv/m/n/10 (ybvr8t) 1
  ....\glue 0.0 plus 1.0fil
+ {8r.enc}{supp-ybv.enc}
  ============================================================
  ============================================================
  TEST 2: test-T1-ybv-m-it

@cfr Er, I guess if the .enc files are not available but skipping them does not lead to a fatal error?

@JosephWright there's no obvious error in the non-normalised log. and I think it should find the .enc files in testdir.

I only seem to get the difference once e.g. cfr-lm gives me

*** ../build/test/cfr-lm-auto-test.tlg  2024-11-11 04:20:17.711867048 +0000
--- ../build/test/cfr-lm-auto-test.pdftex.log   2024-11-11 04:33:59.411866310 +0000
***************
*** 3539,3545 ****
  ....\glue 0.0 plus 1.0fil
  ....\T1/clm2j/m/n/10 (clmr2j8t10) 1
  ....\glue 0.0 plus 1.0fil
! {t1-clm.enc}{dotdigits-clm.enc}
  ============================================================
  ============================================================
  TEST 2: test-T1-clm-m-sl
--- 3539,3545 ----
  ....\glue 0.0 plus 1.0fil

@DavidCarlisle no news on the font issue of my colleague, I guess?

@mickep no sorry I need to re-enable my minion pro stuff so I couldn't just run the test as is, and then I have been a bit distracted with breaking fixing amsmath etc...

@mickep but if setting localmathalphabets to 0 fixes it for you then that will be the answer until the next release, as any change will likely be delicate \global or not changes throughout the font allocation setup and not something you can easily "pach" just disable the mechanism altogether, but the test file will be useful to check the new code if we ever get round to writing it

@DavidCarlisle fixing, breaking, what is the difference? :)

@DavidCarlisle Thanks, I will pass that to him. He was using the latest (a new installation of) miktex, if that can help in any way.

@yo' Ok, thank you for the information

@mickep no the localalphabet feature was added a couple of years ago and it mostly works well but there is some nested math use that throws it out (and once it loses control over which font is assigned to which \fam all hope is lost) so it doesn't depend on miktex/texlive, or the specific fontsreally just specific font packages and particular tricky nested math constructions,

texdev.net/2024/11/11/the-mythical-latex3

@JosephWright LaTeX1e even better than this 2e thing we have been stuck with?

@DavidCarlisle Sounds a bit fragile, but let us hope for the best!

@JosephWright might not eb something you’ll every say you use -> might not be something you’ll ever say you use

@JosephWright latex1e is a nice invention ;-) typos: top pf, ued and at the end eb->be

@JosephWright @UlrikeFischer was faster

@Skillmon and @DavidCarlisle was even faster

@JosephWright youtube.com/watch?v=i2q0T7QXETs s/Scots/L3/g :)

@DavidCarlisle :)

@UlrikeFischer Thanks

@DavidCarlisle Thanks

@DavidCarlisle Thanks

@JosephWright as far as I can tell, it is logging difference caused by one or more .enc files being in testdir rather than texmf-dist. so I'm just going to update the .tlgs, but don't know if l3build wants to document this somewhere?

@cfr Oh, yes, that will do it - path normalisation and all that

@JosephWright maybe, but it doesn't affect the test results for anything else because the differences don't show up inside the test bit. e.g. the paths are different for article.cls if checksearch is false, but it doesn't change the .tlg. neither does moving the .pfbs etc. the .log is different, but the normalised one isn't.