Fix #3614631 by delegating to script modes, instead of just importing their rules.
The indentation rules for the script mode are only applied if (real) delegation is
used. (Some relevant code can be found in JEditBuffer.getIndentRules)
This patch also shortens and simplifies the file nicely, but for the price of reintroducing
a corner-case bug: Closing </script> tags may be consumed by script comments, which
is inconsistent with the XML plugin and most browsers.
Given that it seems impossible to fix this (in a mode file) while using (real) delegation,
I'd say that having correct script mode properties is preferable to not having this
bug.
(Thanks to Matthieu Casanova for pointing in the right direction.)
| Submitted | marchaefner - 2013-07-13 - 19:00:20z | Assigned | daleanson |
|---|---|---|---|
| Priority | 5 | Category | None |
| Status | Open | Group | None |
| Resolution | None | Visibility | No |
| 2013-11-08 - 13:05:52z marchaefner |
Added new version, that also includes a fix for #3615189. (Also renames STRING ruleset to ATTRIBUTE and removes an outdated comment.) |
|---|---|
| 2013-11-11 - 23:07:38z daleanson |
Marc, could you post an example of "Closing </script> tags may be consumed by script
comments, which is inconsistent with the XML plugin and most browsers." ? This patch is looking good to me, and as far as I can tell, even putting a javascript comment in front of the closing script tag looks good. |
| 2013-11-12 - 17:35:18z marchaefner |
Any plain old <script> // comment </script> consumes the closing tag (because the comment is matched by an EOL_SPAN). Subsequent text is still highlighted with JavaScript mode, but a look at the SideKick or a test with a browser reveals that it's actually HTML. (Fixing this was one of the goals of #3595104, but it seems to have caused more trouble than good.) |
| 2013-11-26 - 19:23:31z kerik-sf |
Hi, looks good to me. What about marking the whole end of attribute invalid when invalid entity is detected ? replace <SEQ>...</SEQ> with <SEQ_REGEXP TYPE="INVALID">&[^"]*</SEQ_REGEXP> Otherwise it's rather easy to miss those red & in pink url. Also, the old html mode will show all the javascript code as comment when viewing the source of this page: http://www.cs.umd.edu/~bederson/papers/index.html which the new mode doesn't, which is good! |
| 2013-11-27 - 13:36:18z marchaefner |
> ... it's rather easy to miss those red & in pink url. Yes that is an issue, but I think that would be better addressed by changing the default colors. My motivation are users who are forced to work with broken, but nonetheless perfectly well displayed, HTML files. And I strongly suspect HTML5 makes such a scenario even more likely (it "legalizes" some of those broken cases, to be exact). An overly aggressive error display wouldn't be helpful to those users but a serious distraction. Even more so if they don't use the default color schema (for example my INVALIDs are _really_ red.) |
| 2013-07-13 - 19:33:15z marchaefner |
html.xml.patch HTML mode: Fix bug 3614631 - mode properties in script blocks |
|---|---|
| 2013-11-08 - 13:06:15z marchaefner |
html.xml.V2.patch Fixes 3614631, 3615189, 3615174 |