Thread (1 message) 1 message, 1 author, 2020-10-07

Re: [PATCH v2] userdiff: permit Rust's macro_rules! keyword in hunk headers

From: Junio C Hamano <hidden>
Date: 2020-10-07 06:10:29

"Konrad Borowski via GitGitGadget" [off-list ref] writes:
From: Konrad Borowski <redacted>

Changes since v1:
- Changed macro_rules! to be considered to use the same
  rule as rest of keywords to reduce the size of a change
  as suggested by Phillip Wood. This means that
  'pub macro_rules!` (a syntax error) is considered to be
  a hunk header.

- Written commit message in imperative mood as suggested by
  Johannes Sixt.
The above is where you write the proposed log message, not where you
write the differences between v1 and v2 (which, if exists, would be
below the three-dash line after your sign off).  After all, people
who are reading the final history of the project in "git log" will
not even know or care that what they are reading is the second
iteration and/or there was a first iteration that was different from
the final version.
Subject: [PATCH v2] userdiff: permit Rust's macro_rules! keyword in hunk headers
Somehow I find the "permit" a bit odd---it is not like we are
disallowing Rust programs written in certain way.

     userdiff: recognize 'macro_rules!' as starting a Rust function block

or something?
Signed-off-by: Konrad Borowski <redacted>
---
...
As to the simplified rule that matches 'pub macro_rules!'.

We assume what is written by the end-user, in which these patterns
are used to find function headers, is a syntactically correct
program.  This allows us to write a pattern that may match with a
syntactic nonsense (i.e. false positive is perfectly fine).  What
we care about is to avoid false negatives---if our pattern does not
recognize what users would consider a valid and common function
header line, that is a problem.

So, from that point of view, you do not have to worry too much about
'pub macro_rules!'  matching your pattern.
quoted hunk
diff --git a/t/t4018/rust-macro-rules b/t/t4018/rust-macro-rules
new file mode 100644
index 0000000000..ec610c5b62
--- /dev/null
+++ b/t/t4018/rust-macro-rules
@@ -0,0 +1,6 @@
+macro_rules! RIGHT {
+    () => {
+        // a comment
+        let x = ChangeMe;
+    };
+}
diff --git a/userdiff.c b/userdiff.c
index fde02f225b..c0f63c57b2 100644
--- a/userdiff.c
+++ b/userdiff.c
@@ -165,7 +165,7 @@ PATTERNS("ruby", "^[ \t]*((class|module|def)[ \t].*)$",
 	 "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+|\\?(\\\\C-)?(\\\\M-)?."
 	 "|//=?|[-+*/<>%&^|=!]=|<<=?|>>=?|===|\\.{1,3}|::|[!=]~"),
 PATTERNS("rust",
-	 "^[\t ]*((pub(\\([^\\)]+\\))?[\t ]+)?((async|const|unsafe|extern([\t ]+\"[^\"]+\"))[\t ]+)?(struct|enum|union|mod|trait|fn|impl)[< \t]+[^;]*)$",
+	 "^[\t ]*((pub(\\([^\\)]+\\))?[\t ]+)?((async|const|unsafe|extern([\t ]+\"[^\"]+\"))[\t ]+)?(struct|enum|union|mod|trait|fn|impl|macro_rules[\t ]*!)[< \t]+[^;]*)$",
I notice that you still allow "macro_rules    !" as Phillip pointed
out.  Is that intended?

Thanks.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help