Mike Gerdts wrote:
...
> - How compressible is bytecode compared to uncompiled code?  That is,
> once it is put into a compressed disk image format, how much space is
> really being saved?
> 

All of Mike's points are well-taken, especially this one.  An argument 
that "it might help the Live CD" without data to back it up is not going 
anywhere.  Essentially all of the contents of the live CD are compressed 
in one way or another already, be it by lofi (using lzma) for the major 
parts that stay on the media, or gzip (or dcfs) for the root archive, so 
proposing to compress them by other means is relatively unlikely to lead 
to actual observed benefit.

I'll further add that, in at least some instances, we've actually seen 
adverse impacts based on some of the recent projects.  For example, the 
recent conversion of cut(1) and other objects to ksh93 built-ins 
actually increased the size of the ramdisk required on SPARC by a couple 
of megabytes.  Whether its other benefits outweigh this cost is 
something to contemplate.

At this point, I believe the decision to close this bug was absolutely 
correct.  Presenting contradictory data, not opinions, is the way to 
move this forward.

BTW, if anyone wants to propose ideas for how to save space on the live 
CD or memory in its execution environment, I really encourage engaging 
directly with me and my team (caiman-discuss at opensolaris.org); unless 
you are well-versed in how the live CD is constructed, any assumptions 
are likely wrong.

Dave

Reply via email to