Markup agnostic extra files #1200
Conversation
|
There are definitely concerns about breaking changes here. I realize that you show a non-breaking option, but I'm not 100% sure that dc5788a is really the only breaking change. It seems to me that all of these parser changes would break examples where, for example, a user was attempting to render an HTML document, i.e.: This would potentially incorrectly render the comment in the document, depending on its contents. This is quite an edge case, I realize, but it should be clear that this is technically a breaking change. I actually think that the "2 newlines" option is actually the least breaking, since newlines at the top of the file are undefined behavior. Not sure if that specific heuristic is useful to your use case though. |
Rewrite how comments are handled in textual files (extra objects). Prefer markup-agnosticism over sticking to HTML comments. Do not remove comment tags, as they are bound to specific markup, and hidden from user's eye anyway. Also, see discussion in: https://groups.google.com/forum/#!topic/yardoc/Y700TGceoII
Do not interpret attributes section of an extra file if it starts with two blank lines (previously: one blank line). This is a breaking change, however not very drastical. Seems to be a good compromise between simplicity and ease of use.
skalee
force-pushed
the
markup-agnostic-extra-files
branch
from
dc5788a
to
bcfb36c
Compare
|
Rebased for maintenance. Nothing has changed in commit contents. |
lsegal
force-pushed
the
master
branch
2 times, most recently
from
44c07a8
to
112a890
Compare
Description
Rewrite how comments are handled in textual files (extra objects). Prefer markup-agnosticism over sticking to HTML comments. Do not remove comment tags, as they are bound to specific markup, and hidden from user's eye anyway. As an effect, more types of comments are now supported, including ones from AsciiDoc and Textile. Fixes #1192.
This is a somewhat breaking change (see).
Possible improvements
If required, I can restore previous behaviour changed in dc5788a, making this change a non-breaking one. Well, at least in terms of how it was formally specified in specs.
It's just a matter of adding following lines to the altered
caseconstruct:See also
Completed Tasks
bundle exec rakelocally (if code is attached to PR).