Thread: Postgres code quality
Kudos to developers and community http://lwn.net/Articles/483768/rss The headline is about Linux, but the article and report also covers PHP and Postgres. Postgres code shows well. -- Adrian Klaver adrian.klaver@gmail.com
Excerpts from Adrian Klaver's message of vie feb 24 20:42:49 -0300 2012: > > Kudos to developers and community > > http://lwn.net/Articles/483768/rss > > The headline is about Linux, but the article and report also covers PHP and > Postgres. Postgres code shows well. Meh. As far as I know we still haven't configured the coverity tools so that they don't report spurious errors in certain safe coding idioms we use, so our true defect density is likely to be lower than the article reports. -- Álvaro Herrera <alvherre@commandprompt.com> The PostgreSQL Company - Command Prompt, Inc. PostgreSQL Replication, Consulting, Custom Development, 24x7 support
On Friday, February 24, 2012 4:25:42 pm Alvaro Herrera wrote: > > Meh. As far as I know we still haven't configured the coverity tools so > that they don't report spurious errors in certain safe coding idioms we > use, so our true defect density is likely to be lower than the article > reports. To a certain extent the validity of test results is in the eye of the beholder:) What I found important is that relatively speaking Postgres holds its own and that Postgres was held up as project worthy of comparison. Pushes the project a little more into the limelight. -- Adrian Klaver adrian.klaver@gmail.com
Alvaro Herrera wrote: > As far as I know we still haven't configured the coverity tools > so that they don't report spurious errors in certain safe coding idioms > we use, so our true defect density is likely to be lower than the article > reports. Actually, you'll be happy to hear that before this year's report and as part of moving PostgreSQL to the latest version of analysis, we tuned analysis results. Among other things, the analysis now understands the various elog/ ereport(ERROR/PANIC) idioms which got rid of a lot of false positives, and we also excluded generated code from analysis results. I would expect the results to be quite relevant at this point - if there is additional tuning to be done, please let us know at scan-admin@coverity.com Asya Kamsky Coverity, Inc. -- View this message in context: http://postgresql.1045698.n5.nabble.com/Postgres-code-quality-tp5514252p5516218.html Sent from the PostgreSQL - advocacy mailing list archive at Nabble.com.