Re: GSoC 2008 dma enhancements

Previous thread: link(2) behaviour when creating a hardlink to a symlink by Jordan Gordeev on Wednesday, June 11, 2008 - 4:27 pm. (3 messages)

Next thread: HAMMER update 11-June-2008 by Matthew Dillon on Wednesday, June 11, 2008 - 8:27 pm. (4 messages)
To: <kernel@...>
Date: Wednesday, June 11, 2008 - 8:08 pm

Lets not turn this into a bikeshed that I am then forced to terminate!

In anycase, old-time DragonFly developers will remember when we started
ripping out all uses of sprintf(), strcpy, and the like from the kernel,
and replaced them with e.g. snprintf().

My personal viewpoint is that one should always code defensively, no
matter what the language. And I'm not just talking about buffer overflow
issues, I've seen people get into trouble many times with Java by
chosing horrible non-scaling algorithms which blow out available memory
(as one example).

In the case of C, its a combination of avoiding certain very dangerous
functions like strcpy(), and asserting non-obvious things that should be
true to catch coding bugs.

For a laugh, grep out the number of KKASSERT lines I have in the HAMMER
filesystem source code.

-Matt

Previous thread: link(2) behaviour when creating a hardlink to a symlink by Jordan Gordeev on Wednesday, June 11, 2008 - 4:27 pm. (3 messages)

Next thread: HAMMER update 11-June-2008 by Matthew Dillon on Wednesday, June 11, 2008 - 8:27 pm. (4 messages)