On 20 Jun 2014, at 01:55, Mark van Gulik <[email protected]> wrote:

> So... There's no way we're going to reach our July 1 milestone.

no worries. the gods probably won't notice the slip.

> Todd has built a heck of a lot of Avail features over the last couple of 
> months (git pull will show you what he's been up to), but I've been stalled 
> on file I/O.

i’ve seen the works in the making - it’s cool.

> The main contrivance that I finally settled on was a concealed cache plus a 
> simplified version of a modern cache coherence protocol.  This should make it 
> very efficient to go forwards or backwards in a file (usually without 
> transferring data from the OS), or even read from multiple hot regions of the 
> same file, since the cache may be populated at any time between explicit 
> Refreshes (invalidation of all buffers cached for a file) and Flushes 
> (passing the data to the OS to eventually write to disk).  The cache is tied 
> to a file *handle*, so if only one handle is open it will always be 
> consistent with disk, or at least what disk will look like after all pending 
> writes complete (which you can block for with the Synchronize method).

i didn't want to say it, but there: 

your io library sounds absolutely cool, but it is not avail *specific*: it 
could be implemented in java or another language.
i would rather see some compiler improvements - but that’s just me.

> I'm expecting my day job (and Todd's) to be chomping some extra hours over 
> the next few weeks.  It already is.  So file I/O itself probably won't be 
> ready by July 1, let alone macros.

yah, i know what you’re saying. again, don’t worry.
io is cool. macros just as much. it’s just a matter of priorities: after all, 
day jobs pay the bills - we have families to care for.

… oops. now that you mention it: 
spread version v0.2 is still not done! 

> Hopefully this look behind the scenes will let interested parties see that 
> missing the July milestone is the furthest thing from being a sign that Avail 
> is becoming a defunct project!

defunct? after 25 years of hard work? nah, we don’t believe you!

cheers,
Robbert.

Reply via email to