svn commit: r230354 - head/usr.sbin/makefs

[Eitan Adler]

On Fri, Jan 20, 2012 at 7:13 PM, Bruce Evans <brde at optusnet.com.au> wrote:
> I agree.  Never fix vendor code.  Especially style bugs in it.  Not all
> vendor code is in contrib.

I'm sure this is a silly question, but why isn't this utility in a
contrib/ folder?

> In fact, it seems that there there was a very large amount of unused code,
> and this commit only removes the tip of it.  There is lots more code to
> initialize msrargs.data in the above, but now msrargs.data is unised too.
> This is apparently too complicated for the compiler to see that it is
> unused:

I had more patches which this conversation is making me hold back on:
this code is contributed it doesn't fall under my goal of making the
world buildable with gcc46.

[snip other fixes to the api]

> I don't understand this code well enough to fix it.  Fixing it requires
> understanding whether the unused variable was unused because of another
> bug.  Another bug seems likely here, since there is so much dead code.
> Or maybe I'm just confused, and the dead code is actually undead.  It
> is complicated enough for this to be unclear.

I did not just rely on the compiler - I went through the code manually
to ensure it was correct.

-- 
Eitan Adler
Ports committer
X11, Bugbusting teams