A few fixes, a metric f-ton of tests and CI#43
Conversation
|
@sgrif don't review the tests, it's AI, just the changes to |
|
Sorry, I only accept imperial fuck-tons |
| } | ||
| } | ||
|
|
||
| pub fn run_parse_walk_case(query: &str) { |
There was a problem hiding this comment.
Walking is unnecessary, just debug printing the root is enough.
There was a problem hiding this comment.
I thought you were worried about that one or something because it did something different. We can parse all DDL but we can't walk it for example, so I was just making sure it can walk DML
|
Honestly, seeing this is making me feel that we should bring in the build changes but not the 358388627287 tests. Especially the giant comments documenting the PG grammar which will get out of date feel very off to me. I'm happy this exercised all the obscure ddl statements and found things but I'm not particularly concerned about regressions there. Does that make sense? |
|
Yup |
|
Actually nope. Because of comments. They can change them and we won't know, we need to exercise the parser on as much grammar as we can. Its not type checked until we parse basically |
|
I don't see that as a realistic concern. Every problem case has
specifically been that the comments *aren't* getting updated to keep up
with grammar changes, and these tests won't help at all with that.
I'm also wondering if we should just yeet the comment typing and instead
allowlist the ones we want
…On Thu, Jul 16, 2026, 6:44 PM Lev Kokotov ***@***.***> wrote:
*levkk* left a comment (pgdogdev/pg_raw_parse#43)
<#43 (comment)>
Actually nope. Because of comments. They can change them and we won't
know, we need to exercise the parser on as much grammar as we can
—
Reply to this email directly, view it on GitHub
<#43?email_source=notifications&email_token=AALVMK37O2QXBY32P4MAISL5FFZILA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOJZG44DONZRHE2KM4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#issuecomment-4997877194>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AALVMKZE2KIE6KJ4S7343V35FFZILAVCNFSNUABGKJSXA33TNF2G64TZHMYTENZRGU3TQNJXGI5US43TOVSTWNBZGA3DQNZQGE3DPILWAI>
.
Triage notifications, keep track of coding agent tasks and review pull
requests on the go with GitHub Mobile for iOS
<https://github.com/notifications/mobile/ios/AALVMK75FC7OJOFSE7UEM2D5FFZILA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOJZG44DONZRHE2KM4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJKTGN5XXIZLSL5UW64Y>
and Android
<https://github.com/notifications/mobile/android/AALVMKZXDAMNGUDN6E67OLD5FFZILA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOJZG44DONZRHE2KM4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJLTGN5XXIZLSL5QW4ZDSN5UWI>.
Download it today!
You are receiving this because you modified the open/close state.Message
ID: ***@***.***>
|
|
Yeah food for thought. I dunno, I feel like keeping the tests is free, removing them is not. If they never fail, cool, but if they do, we might catch something. What's the downside? Pg syntax only adds right, so they will never create noise unless it's real |
|
Can we at least remove the comments that will end up quickly getting out of
date?
…On Thu, Jul 16, 2026, 6:51 PM Lev Kokotov ***@***.***> wrote:
*levkk* left a comment (pgdogdev/pg_raw_parse#43)
<#43 (comment)>
Yeah food for thought. I dunno, I feel like keeping the tests is free,
removing them is not. If they never fail, cool, but if they do, we might
catch something. What's the downside? Pg syntax only adds right, so they
will never create noise unless it's real
—
Reply to this email directly, view it on GitHub
<#43?email_source=notifications&email_token=AALVMK76GSD5S3WK2CYYKBD5FF2AFA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOJZG44TCMZRGU32M4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#issuecomment-4997913157>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AALVMK2PCVCPETFYWU65DSD5FF2AFAVCNFSNUABGKJSXA33TNF2G64TZHMYTENZRGU3TQNJXGI5US43TOVSTWNBZGA3DQNZQGE3DPILWAI>
.
Triage notifications, keep track of coding agent tasks and review pull
requests on the go with GitHub Mobile for iOS
<https://github.com/notifications/mobile/ios/AALVMK6FROG3J6SOJ2J7CNT5FF2AFA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOJZG44TCMZRGU32M4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJKTGN5XXIZLSL5UW64Y>
and Android
<https://github.com/notifications/mobile/android/AALVMK64I23HFEYEVJ76AYD5FF2AFA5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIOJZG44TCMZRGU32M4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJLTGN5XXIZLSL5QW4ZDSN5UWI>.
Download it today!
You are receiving this because you modified the open/close state.Message
ID: ***@***.***>
|
|
Mouahaha, green stars up. Now you gotta delete 5k 📈 |
|
Ah, I got it. Yeah I'm not worried about it handling dml. It's documented
…On Thu, Jul 16, 2026, 9:58 PM Lev Kokotov ***@***.***> wrote:
***@***.**** commented on this pull request.
------------------------------
In tests/common/mod.rs
<#43 (comment)>:
> +#![allow(dead_code)]
+
+use pg_raw_parse::{parse, walk::walk};
+
+pub fn run_parse_debug_case(query: &str) {
+ let ast = parse(query).unwrap_or_else(|err| panic!("failed to parse `{query}`: {err:?}"));
+ assert!(!format!("{:?}", ast).is_empty())
+}
+
+pub fn run_parse_debug_cases(cases: &[&str]) {
+ for query in cases {
+ run_parse_debug_case(query);
+ }
+}
+
+pub fn run_parse_walk_case(query: &str) {
I thought you were worried about that one or something because it did
something different. We can parse all DDL but we can't walk it for example,
so I was just making sure it can walk DML
—
Reply to this email directly, view it on GitHub
<#43?email_source=notifications&email_token=AALVMK7OVJIRXGQ7RSZ4K7T5FGP7XA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTINZRHEZDGOJUGEY2M4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#discussion_r3600275494>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AALVMK2L6EFME4TFGTA3YKL5FGP7XAVCNFSNUABGKJSXA33TNF2G64TZHMYTENZRGU3TQNJXGI5US43TOVSTWNBZGA3DQNZQGE3DPILWAI>
.
Triage notifications, keep track of coding agent tasks and review pull
requests on the go with GitHub Mobile for iOS
<https://github.com/notifications/mobile/ios/AALVMKZF2M6EGXMECTGOIFD5FGP7XA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTINZRHEZDGOJUGEY2M4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJKTGN5XXIZLSL5UW64Y>
and Android
<https://github.com/notifications/mobile/android/AALVMK6EXDMLAAULF5376E35FGP7XA5CNFSNUABKM5UWIORPF5TWS5BNNB2WEL2QOVWGYUTFOF2WK43UKJSXM2LFO4XTINZRHEZDGOJUGEY2M4TFMFZW63VMON2GC5DFL5RWQYLOM5S2KZLWMVXHJLTGN5XXIZLSL5QW4ZDSN5UWI>.
Download it today!
You are receiving this because you modified the open/close state.Message
ID: ***@***.***>
|
CREATE TABLE ... PARTITION BY LIST