Test UTF-8 paths on AppVeyor#1449
Conversation
|
Just out of curiosity:
+ cp $f "$OCAMLROOT/bin/flexdll/$f"
Why do you repeat $f?
I'd write it:
+ cp $f "$OCAMLROOT/bin/flexdll"
What do you think?
|
|
I wouldn't type it in a shell, but I generally prefer to be explicit in a script. That said, I would always (except when I forget!) append a slash in your version, to ensure that the files don't get created as flexdll if for some reason the directory doesn't exist. |
|
David Allsopp (2017/10/25 13:42 -0700):
I wouldn't type it in a shell, but I generally prefer to be explicit
in a script. That said, I would always (except when I forgot!) append
a slash in your version, to ensure that the files don't get created as
flexdll if for some reason the directory doesn't exist.
/ would be fine.
|
|
Yes, I wouldn't mind merging this in 4.06 given that it only affects the CI (we can easily correct it after the release, and it's fine if it doesn't get in user tarballs as they don't invoke these scripts themselves on a non-development version). I'm using "consider-for-4.06.0" because it has the same semantics to me or Damien as "4.06.0", and it more clearly suggests to others that things may still be left out if time runs out. I'm not sure why the CI fails here, and it's not easy to see which changes are not in the PR anymore since #1447 was merged -- maybe a rebase on your end could help. |
7a770bf to
34ea18b
Compare
|
It was failing because #1448 isn't merged. The changes to Readme.win32.adoc hadn't ended up in the wrong GPR - this GPR was (intentionally) based on #1447. I've now rebased it on to #1448 so while it still shouldn't be merged, it's just possible that AppVeyor will pass, unless I messed up the rebase! |
I notice the subtle way you complain about having to split PRs, but I am not moved :-) |
|
Don't accuse me of subtlety - I prefer sledgehammers! 😄 Actually, what I referring to was rebasing on top of #1287 which included some nice tidying of the AppVeyor script... so nice, that I do appear to have cocked up the rebase of this one! |
34ea18b to
7289bcf
Compare
|
Let's see how this one fares |
This was set before running the testsuite, but it's not necessary (the testsuite contains its own trickery to ensure that CAML_LD_LIBRARY_PATH is correctly set).
All three builds on AppVeyor updated so that the build takes place and has PREFIX set to a directory which includes 2+ bytes UTF-8 sequences and UTF-16 surrogate characters.
7289bcf to
63e1460
Compare
|
The prerequisites of this GPR are now met ... assuming it still passes CI! |
|
Edit: the message below was already-outdated, keeping it for posterity:
|
Test UTF-8 paths on AppVeyor (cherry picked from commit c281fae)
|
Cherry-picked in 4.06 : b975103. |
See #1444.
This is based on #1447, just so that the diff is unaltered; it's also dependent on #1446 and #1448