Vote: What level of automatic code highlighting do we want?

Question

Now that have code highlighting enabled, we need to determine how 'automatic' it should be.

The code highlighting engine works in two ways:

1. Automatic inference of the language in use via the tags on the question, and
2. Manual specification of the language in a code block, overriding any language behavior on a question tag

Currently, we have highlighting enabled for a large number of tags, a default set from Stack Overflow or some such. We've already disabled automatic highlighting on a small handful (ios, iphone, php) due to awkward highlighting on questions, but we need to decide what approach to take for a consistent behavior across all tags.

Answer 1

Enable it extremely selectively on tags where the content of a code block has a high chance of actually being code.

For instance, enable these:

• lang-sh - bash sh shell-scripting
• lang-vb - vbscript

But, leave most everything else disabled.

• Our mysql postgresql sql-server questions should be more configuration and errors than SQL syntax; lang-sql would be more detrimental then helpful.
• Our python django questions are rarely code, and rarely even stack traces; lang-py would be more detrimental than helpful.
• Our php php5 questions are rarely code, lang-php would be more detrimental than helpful.
• ...etc...

Answer 2

Disable all automatic highlighting, globally, across all tags.

You won't see a lot of highlighting at all, since it needs to be manually specified by the author (it's explained in editing help, but to really do it right takes a dig in m.SO) - but at least there won't be any highlighting of something that isn't code.

Answer 3

Enable it on all tags for which there's a language available, even if we get some false-positive highlighting.

The increased incidence of highlighting on actual code is worth it, and/or the coloring is subtle enough that it doesn't bother me.