Hi Scott, Scott Stanton wrote: > Basically, the way perlapp works seems to address all of our concerns, > so I'd like to see if we can get equivalent functionality into par. I'm > not suggesting abandoning other modes of operation, just adding the > ability to use a per-user shared cache to avoid most of the overhead.
I agree that this would be desirable. However, there are various different ways in which stuff can get extracted already: 1) The MD5'd bootstrap files. 2) full extraction from binaries (inc) 3) "use PAR 'foo.par';" works somewhat differently again, if I recall correctly. 4) dll's are extracted to shlib 5) Using --clean, some things change So this would add a sixth layout or extraction component. And I'm sure I actually forgot something, too. Some of these scenarios are actually combined in a single use case. For example, a pp'd binary executable can "use PAR 'foo.par'". What would happen for this with the proposed layout? What would be the default way of caching? How would the two caches mix? Perhaps it would be the right time to refactor the code that does the extraction to something more abstract? Best regards, Steffen
