CRLF preservation - Githubissues

dra27 commented 6 years ago

This PR makes the following changes:

OpamLexer requires some additional state which allows it to track the line-endings of the file being parsed.
Based on this state, OpamBaseParser's EOF token is augmented with a bool option indicating whether the file had CRLF-encoded line endings, LF-encoded line endings or a mixture.
OpamParser then treats mixed endings as CRLF on Windows and LF elsewhere and returns an opamfile record augmented with a file_crlf field.
OpamPrinter uses this information to control how it converts opamfile and related values back to strings.

Important notes:

The Normalise functions are unchanged - in particular, they use LF-endings on Windows
The Preserved functions now no longer LF-endings for changed entries and CRLF-endings for any unchanged values in a file which was CRLF-encoded
opam format upgrade should not magically convert all CRLF-encoded files to LF-encoded files.

This is still partially a work-in-progress, but is ready for initial review - in particular, I plan to add CI to this repo and some tests of these various behaviours and also produce a parallel PR in opam itself, as some changes are required there for CRLF-preservation.

In addition, the API is now documented.

dra27 commented 6 years ago

https://github.com/ocaml/opam-file-format/pull/11/commits/3632a6ded471681347efd663a55faaf8c4969275 contains an unrelated, but important, change to the encoding of strings containing newlines.

dra27 commented 6 years ago

Tests to follow - @AltGr, is AppVeyor enabled as well?

avsm commented 5 years ago

just before i rebase #13 -- is this mergeable now?

rjbou commented 4 years ago

Thanks @dra27! I ported half of the PR in #21 for CI & #22 for doc. I'll ping @altgr for CLRF review

avsm commented 4 years ago

Dev meeting discussion: this is for opam 2.2 at the earlier

ocaml / opam-file-format

CRLF preservation #11