changes to library files can still bleed through subtly

[ahl]

Let me start by saying this isn't a big deal.

#85 (thank you!) removed $WORKSPACE-related line numbers, but the output can still change spuriously when library source files cross ⌊log_10⌋ boundaries. For example, in the output below, the file handler.rs just went over 1000 lines:

-   --> $DIR/bad_endpoint6.rs:22:8
-    |
-22  |     Ok(HttpResponseOk(Ret { "Oxide".to_string(), 0x1de }))
-    |        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `serde::ser::Serialize` is not implemented for `Ret`
-    |
-   ::: $WORKSPACE/dropshot/src/handler.rs
-    |
-    | pub struct HttpResponseOk<T: JsonSchema + Serialize + Send + Sync + 'static>(
-    |                                           --------- required by this bound in `dropshot::handler::HttpResponseOk`
+    --> $DIR/bad_endpoint6.rs:22:8
+     |
+22   |     Ok(HttpResponseOk(Ret { "Oxide".to_string(), 0x1de }))
+     |        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ the trait `serde::ser::Serialize` is not implemented for `Ret`
+     |
+    ::: $WORKSPACE/dropshot/src/handler.rs
+     |
+     | pub struct HttpResponseOk<T: JsonSchema + Serialize + Send + Sync + 'static>(
+     |                                           --------- required by this bound in `dropshot::handler::HttpResponseOk`

The only way that occurs to me to fix this would require multiple passes. I think the simplest option would be to post-process looking to see if all lines match ([ ]*-->|\d*[ ]*[ ]+ |[ ]*:::).*. In that case one could trim one or more ' 's from each line.

Happy to submit a PR 1. this is worth fixing and 2. that approach sounds reasonable.

[dtolnay]

I would accept a PR to fix this. Around above this line you'd be able to scan forward whether every line until the next blank line matches your pattern, and if so, decide how many spaces to be removed. Save that in a field of Filter and every subsequent line could delete the first n spaces from the line, until reset to 0 by a blank line.