Conversation
zng/docs/spec.md
Outdated
There was a problem hiding this comment.
| > For example, the zqd server uses application-defined message `0xf7` | |
| > For example, the [zqd](../../cmd/zqd) server uses application-defined message `0xf7` |
mccanne
force-pushed
the
zng-beta
branch
2 times, most recently
from
747d33b to
b4f6a5a
Compare
expr/functions.go
Outdated
There was a problem hiding this comment.
| "typeof": {1, 1, typeOf}, | |
| "iserr": {1, 1, isErr}, | |
| "typeOf": {1, 1, typeOf}, | |
| "isErr": {1, 1, isErr}, |
There was a problem hiding this comment.
@philrz and I talked about function names and I think we need a group consensus around naming conventions. The current name all are package-style but we don't have packages so that seems wrong. I'm not sure about lower camelCase. It seems a little arbitrary. Also for a interactive query language where a user often type lots of stuff rapid fire, getting capitalization right seems like a mistake. I guess I am arguing for snake case where underscores are used only for long names that are less often types. How about I leave this for now and we do a PR soon on whatever we decide. Issue #1426
zio/ndjsonio/reader.go
Outdated
There was a problem hiding this comment.
We should add a comment here (or above the zenum alias) that these are expected to be temporary pending changes in json ingest handling, and were here for compatibility just for the existing json ingest process.
There was a problem hiding this comment.
There don't seem to be any tests that exercise this. If I comment them out, all tests pass. But I'm pretty sure the typings stuff needs this and you are totally correct that the typings schema policy should define what zenum and port are rather than being hardcoded here. I will add the comment you suggest. I can't find an issue where to hang this so I will create a new one but it will probably be redundant with the larger task of imrpoving schema maps. #1427
zio/tzngio/writer.go
Outdated
There was a problem hiding this comment.
Is there a github issue you can reference here?
zio/tzngio/writer.go
Outdated
expr/expr_test.go
Outdated
There was a problem hiding this comment.
Remove, or add comment to github issue to re-instate port searching.
There was a problem hiding this comment.
I'll remove the commented-out code. Issue #1428
zng/enum.go
Outdated
There was a problem hiding this comment.
Would you please add a ztest that uses an enum? I didn't see one.
There was a problem hiding this comment.
Oops, good catch. I found the enum test I thought I had committed and it reminded me I had another loose end to take care of... all done.
There was a problem hiding this comment.
BTW, there's some hokiness to the enum values in the type syntax...
#0:record[e:enum[int32,foo:[1],bar:[2],baz:[4]]]
This is the first time value syntax appeared in the type syntax.
I wrapped the values here in [] since this is escaped by the tzng value syntax and commas aren't. It's a bit ugly. I had one thought here: we could change the tzng value encoding to use comma separation instead of semi-colon termination (and otherwise keep brackets) so that the type syntax and value syntax is consistent (@philrz suggested this a year ago). Now would be the time to change this since it will break tzng format (it also messes up the azng updater). The good news is only two small places in the code would need to change (but many tests would need to be updated).
…hilrz This is an auto-generated commit with a zq dependency update. The zq PR brimdata/super#1457, authored by @philrz, has been merged. Link fixes mostly related to ZNG spec update While it's convenient to be able to link directly to anchors for specific sections in docs as we've been doing, one of the hazards is that when a destination changes, the link doesn't show up as "broken" by our automation because the user just ends up being redirected to the top of the page in question. brimdata/super#1394 renumbered some sections in the ZNG spec, so here I've manually rechecked relevant anchor links and made sure they go to the right places again. While I was at it, I found some absolute URLs that I changed to be relative. This makes for better links when working in branches.
This is an auto-generated commit with a zq dependency update. The zq PR brimdata/super#1515, authored by @nwt, has been merged. Handle sets with a complex element type brimdata/super#1394 changed the zng specification to allow sets with a complex element type. This implements the change.
5 participants
This commit updates the ZNG spec based on deployment experience,
further comparisons with other approaches, and team feedback.
We hope to make no more backward incompatible changes and have
the ZNG format finalized by spring 2021.
Various encoding changes were made to improve the ergonomcis
of the format and simplify the implementation.
New types haven been added: bytes, enum, map, type, and error.
Float32 is coming soon in a subsequent PR.
New functions are: typeOf(), isErr(), fromBase64(), and toBase64().
The index operator (i.e., []) now works on records and maps,
in addition to arrays, and record types can now handle arbitrary
strings as keys, e.g., rec[".@_FOO"].
The alpha-zng autodetector now works.
Closes #1291, #1314, #1315, #1317, #1365, #1220
OLD DRAFT COMMENT
Here is a draft for commentary of the zng-beta branch as a draft merge off of the eval branch. This will hopefully get merged shortly after the big eval PR is merged...
Please provide any thoughts or feedback on the beta version of zng/docs/spec.md. The intention is to have no more backward incompatible changes and hopefully also no more forward-incompatible changes but I'm less concerned about the latter.
Since putting up this PR, I updated the zngio reader/writert o conform with the new control codes. Now that this is in, the alpha-zng autodetector should work.