Generate proper precede requirement based on layout follow restriction #9
Conversation
|
In the middle of the night, while in vacation my mind suddenly jumped to: we shouldn't use |
|
Fixed in the subsequent two commits. |
There was a problem hiding this comment.
Hi Davy, I think this PR is now ready for review by you. The commit messages should provide a high-level summary of what changed.
I've tried to isolate the changes in a way that avoids merge conflicts with the other open branches. Notably, there's a new module with functions to compute first/last/precede/follow sets. I reckon other parts of the analysis code can be simplified a bit by using this new module, but let's do that in a separate smaller PR, when the other open branches are merged (I left a TODO in the code so as not to forget).
DavyLandman
changed the title
| ? union(\set, firstOf(t)) | ||
| : ret[delabel(h)]; | ||
| solve (ret) { |
There was a problem hiding this comment.
nit: solve returns the solved value, so this can become: return solve(ret)...
There was a problem hiding this comment.
I played a bit with this just now, but it seems solve doesn't always return a value (it doesn't in the cases of this PR). I submitted an issue basically to update the docs (it seems by design instead of a bug).
|
LGTM, @sungshik |
I noticed that on rascal we were getting stuff like
int aer32 = 34where32was highlighted as a number.I got around that by using the
\bseparator, and rewriting the grammar a bit while doing it. It might be a worthy inspiration, but just parking it here.