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
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 11/22/2013 09:37 AM, Jan Safranek wrote:
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.
I don't know if this is useful to you, but we could try to install the official ReviewBot extension on our ReviewBoard instance:
https://github.com/reviewboard/ReviewBot
(It's an upstream-provided extension to ReviewBoard for running tasks automatically). Upstream-itself uses it to run pep8 checks on all submitted patches, for example.
Reviewbot was renamed to scanbot to avoid confusion with Review Board's own ReviewBot.
Now the scanbot is live and schedules new scans every 10 minutes or so. Red Hat 'donates' the infrastructure and CPU cycles, still the scans run at low priority and it can take several hours on busy days.
Jan
On 11/22/2013 03:37 PM, Jan Safranek wrote:
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 _______________________________________________ openlmi-devel mailing list openlmi-devel@lists.fedorahosted.org https://lists.fedorahosted.org/mailman/listinfo/openlmi-devel
openlmi-devel@lists.fedorahosted.org
-
Jan Safranek -
Stephen Gallagher