Hi,
I'm developing automatic 'reviewbot' tool, which will scan all patches
on our reviewboard. I'm running it manually on selected reviews for now.
Eventually I'll turn it into a cron job and run it automatically for all
reviews. Please report any issues.
The scan leverages tools which we have available internally in Red Hat,
namely clang and Coverity.
The reviewbot:
- Tries to detect sequence of patches by looking at "_prefix_[xx/YY]" in
review 'Summary' and submits whole sequence with the same _prefix_ as
one scan.
- If any item is missing in the sequence, the whole sequence is ignored.
- Otherwise, it performs separate scan for each review.
- The patches under review are tested with current git origin/master.
- It posts "Autoscan started" and "Autoscan finished" comments with
errors introduced by the patches under review. Errors already present in
the code are _not_ reported.
If you submit a series of patches, make sure they have some prefix
before [xx/YY] sequence numbers and do not use [] anywhere else in patch
summary.
Good example: "doc updates [1/10]: Rewrite XYZ"
Bad example: "[1/10]: Rewrite XYZ" -> it MAY clash with other [1/10]
patch on review.
Acceptable example: "[1/1]: Just one patch" -> [1/1] patches do not need
unique prefix.
Jan