Add an appendix with collected ABNF grammar #498
Conversation
| <!-- Update the collected grammar files using `make sourcecode`, which --> | ||
| <!-- is currently manual as it creates a little circular dependency. --> | ||
| <!-- The filenames of the ::includes are likely to change when --> | ||
| <!-- kramdown-rfc-extract-sourcecode handles filenames better. --> |
There was a problem hiding this comment.
With this branch, my workflow breaks. I normally issue scripts/gen.sh, but this now fails with:
Created file draft-ietf-jsonpath-base.txt
Created file draft-ietf-jsonpath-base.html
stdin(212:0): error: Rule jsonpath-query was already defined on line 0 of stdin
stdin(213:0): error: Rule segments was already defined on line 1 of stdin
stdin(214:0): error: Rule root-identifier was already defined on line 4 of stdin
...
stdin(389:0): error: Rule normal-index-selector was already defined on line 207 of stdin
parsing failed: 78 errors encountered
make however succeeds (after I installed npm), so maybe this doesn't matter.
There was a problem hiding this comment.
The problem with duplicating the ABNF is that ABNF doesn't accept duplicated rules (we learned from that in CDDL, which happily accepts [and checks!] duplicated rules).
We could change gen.sh to use sourcecode/abnf as well.
There was a problem hiding this comment.
The problem with duplicating the ABNF is that ABNF doesn't accept duplicated rules (we learned from that in CDDL, which happily accepts [and checks!] duplicated rules).
Fair enough.
We could change gen.sh to use sourcecode/abnf as well.
Yes please.
There was a problem hiding this comment.
The problem with duplicating the ABNF is that ABNF doesn't accept duplicated rules (we learned from that in CDDL, which happily accepts [and checks!] duplicated rules).
Fair enough.
We could change gen.sh to use sourcecode/abnf as well.
Yes please.
Done.
Now we just have to remember to do a make sourcecode manually; you could add this to gen.sh but because of the circular dependency I'm refraining from that now. (We also could put the bap checks in make sourcecode... But then we require people to have bap installed.)
Co-authored-by: Glyn Normington <glyn.normington@gmail.com>
Co-authored-by: Glyn Normington <glyn.normington@gmail.com>
Well, I hope they don't have to do anything, as the ABNF is already fixed. |
| @@ -13,4 +13,5 @@ endif | |||
|
|
|||
|
|
|||
| sourcecode: draft-ietf-jsonpath-base.xml | |||
| rm -rf sourcecode.ba? | |||
There was a problem hiding this comment.
Why does this not remove sourcecode.bak?
There was a problem hiding this comment.
Why does this not remove
sourcecode.bak?
It does. kramdown-rfc-extract-sourcecode then creates a new backup of the existing sourcecode directory.
A bit belt and suspenders, but I wanted to make sure the tool doesn't delete user information.
(We could remove the existing sourcecode directory as well, but then recovery is a bit harder if the sourcecode extraction fails for some reason.)
There was a problem hiding this comment.
Sorry, I didn't express myself very well. Why doesn't the code say the following?
| rm -rf sourcecode.ba? | |
| rm -rf sourcecode.bak |
Are there some other files/directories matching sourcecode.ba? that also need removing?
There was a problem hiding this comment.
Because if the backup mechanism can't create a .bak, it will create .bal, .bam etc.
So if you did an extract-sourcecode manually a few times, you might have all these dropping.
.ba? takes care of all ~15 of them.
BTW I really like this PR and think we should merge it. It's a convenience to certain users of the spec at very little cost to us or readers who are not interested in the collected ABNF. |
|
(Based on an ARTART review comment by Darrel Miller.) |
We don't have to merge this, but this PR shows how to do it.