[syslinux] MEMDISK bug in Syslinux 3.10

[H. Peter Anvin]

Peter Jones wrote:
> 
> I know it wouldn't help this particular case, but it would be nice to
> institute a policy such that:
> 
> 1) the tree isn't considered ok for release until that exact tree has
> been available for testing for some fixed (and announced/posted on the
> website/etc) amount of time, say 2 days or so.  I know I didn't get a
> chance to test -21 this time until right as you shipped final, and I
> suspect I'm not the _only_ one who has trouble keeping up during the
> last moments of "I'd like to ship it soon" panic.
> 

I will generally hold a release if I get an explicit request to do so. 
However, I would prefer to not hold off too long, partially because I 
tend to get time to work in syslinux in a rather bursty manner, and 
there is a very real risk that if I don't get it out, it won't happen.

> 2) the last candidate is what ships, rather than the last candidate plus
> whatever fixes happen.
> 
> That is, instead of "ok, so once this bug is fixed it's safe to ship",
> making that bugfix the criteria for another release candidate, at which
> time the clock resets.  This seems to happen for the second half of the
> release pairs, but not the first.  Is that just my impression?
> 
> (I realize that in the case of 3.08 I was a contributor to this problem,
> and that I should communicate with you more before our releases rather
> than noticing that you're about to do a release and then trying to
> quickly synchronize our time lines.  I'm sorry about that, and I'll try
> to do better.)

In this case, not really, 3.10 is identical to 3.10-pre21.

	-hpa