Archive for June, 2006

Bugbusters

Wednesday, June 28th, 2006

I almost finished my bug buster threemester, fixed dozens small bugs, several serious issues and now trying to find something interesting to do. Actually I still have some not closed PRs in my queue but most of them just awaiting for MFC.

Some thoughts about the whole state of our bug busters team in order of appearance. First of all it is clear the bug busters team is small and we are suffering from that. We need more people working with PRs on regular basis.

Working with PRs you need to learn new code quickly, often change your focus, prepare test cases, do bother with core dumps and live kernel debugs wondering what an author of the code tried to do etc. It is a thankless job because people love new features often assume they are bugless or fixes come for free. Nothing wrong with that, people are just people. But I don’t think anybody would call this work interesting or “fun”. On the other hand it is very important for the project.

With the current incoming PR rate we need two or better three skilled hackers working almost fulltime. Fulltime means money. Perhaps a company which makes money on bug fixing is a good idea but I’m not sure it will survive.

In conslusion several links regarding PR:

gjournal

Monday, June 19th, 2006

pjd’s gjournal looks interesting.

The best of today maillists

Tuesday, June 6th, 2006

Kirk McKusick explains why struct thread passed to all syscalls.

Alex Lyashkov announces his jail project.

Robert Watson about libmemstat(3) tools.

Today goodies

Sunday, June 4th, 2006

Merged to RELENG_6 getent(1) utility and du -n flag implementation.

More fsdb(8) fun

Friday, June 2nd, 2006

OK, just committed fsdb 8) findblk command implementation and made “btime” command – change birth time of inode. Btw, you can see file birth time by a recently added to ls(1) -U flag but no tool to change. Actually, not sure we really need it.

deshallify.sh

Thursday, June 1st, 2006

Never know we have such a great script :-)

One shot - five dead

Thursday, June 1st, 2006

Just fixed long standing /etc/group parsing bug. I was really amused when GNATS shown me five PRs about the same problem.