[ccache] The future of ccache

[John Coiner]

I'll second this question. I maintain a fork of CCache 2.4 internal to 
my workplace, and I'd like to know if there's going to be some active 
work on ccache.

My fork has two changes.

First, it's NFS-safe. It removes the use of non-NFS-safe fcntl() file 
locking and the statistics tracking features that relied on it.

Second, it supports a two level cache. Why? With a one-level cache, you 
have to compromise. You can share the cache among a large number of 
users to get a high hit rate, or you can use hardlinks to conserve disk 
space. You cannot safely do both -- it would be too easy for a 
well-meaning user to change the contents or permissions of their own 
workspace, and in so doing, unwittingly modify the shared cache. 
Basically, a two-level cache allows you to do both safely.

Those aren't for everyone. The NFS fix is interesting for people stuck 
with NFS. The two level cache is interesting for cube farmers stuck with 
any filesystem that doesn't support deduplication. (If you have 
deduplication, you can set up an ordinary one-level shared cache, 
disable hardlinks, and you will have a simple, safe and high performance 
setup that is efficient with disk space.)

So maybe this is best kept as a fork. Anyway it seemed like a good time 
to take a straw poll to see if anyone was interested in those. Anyone?

Cheers,

John



Joel Rosdahl wrote:
> Hi Tridge and others,
> 
> The latest ccache version was released over five years ago. What's the
> future of ccache?
> 
> The reason I'm asking is that I have been hacking on ccache (more about
> that in a separate mail), and, if possible, I would prefer to
> contribute my work to ccache in some way instead of forking my own
> ccache clone project. So, what's the way forward here? If wanted, I'm
> open to help (co-)maintaining ccache.
> 
> Regards,
> Joel
> _______________________________________________
> ccache mailing list
> ccache at lists.samba.org
> https://lists.samba.org/mailman/listinfo/ccache
>