Numbers at end of strings aren't highlighted as part of the string

[caged]

I updated to 1.21 and noticed this started to happen. The digits at the end of the string should be the same color as the string.

language-sql
├── 0.19.0
├── https://github.com/atom/language-sql
├── SQL language support in Atom
├── 120180 downloads
└── 29 stars

[DavidLGoldberg]

Yeah, this is definitely new in 1.21. I'm writing a semantic highlighting package, but utilizing the grammar, and a few of my tests broke :(

Is this going to be fixed soon. I'd rather not have to work around it :-)

[winstliu]

I'll see what I can do today.

[DavidLGoldberg]

@50Wliu, I'd love to take a look tonight and this weekend, but couldn't quickly spot the commit, do you happen to know which one it was that broke it?

[winstliu]

@DavidLGoldberg If I had to guess, #24 (there was a small modification in 05efcd8 though). I think an underscore is a valid word boundary so you're going to have to work around that through a negative lookbehind.

[DavidLGoldberg]

The SHA above is definitely the one that breaks it from what I can tell with:
apm develop language-sql and restoring the previous.

It also passes all of my tests (I thought I introduced another issue, but it turns out it was just that line from the SHA :\ )

I'll try to investigate further. as far as seeing if it potentially breaks something else too? Although, I have to confess I have very little chance of helping out with that regex :(
Where are the best docs/tutorials on this grammar stuff btw? Sublime community still maybe?

[winstliu]

Docs are somewhat lacking on grammar development, unfortunately. In general Atom follows the Textmate guidelines since most of the grammars were converted from Textmate, but the guidelines are pretty vague.
There's an ongoing PR to develop a more cohesive standard in atom/flight-manual.atom.io#114.

[diesire]

I'm having the same problem with v0.20.0

Note that table name T1 is OK, but all fields are broken

I guess it's because T1 is recognized as an entity but FIELD not, so in the first the entity takes precedence over the number but in the last the number rule applies

[winstliu]

Fixed by #34.