The "fast" mode is twice as fast at "slow", which I see quite a few people enjoy (these plugins can be quite slow). Initially I measured the increase in size to be 3-5%, which was why I just flipped default to "fast". It turned out the projects I measured were rather best-case, and a little more experience seems to indicate a 10-15% size increase being more of the norm.
So I have flipped the default back to "slow". Which mode is "best" depends largely on your perspective ;) I'd say fast beats slow any day of the week, but I think 10-15% is a bit too much ;) BTW; The main part of the increase is actually caused by some jars in central having little or no compression applied to them. There might be room for making the compression header sniffing even smarter (recompress if all files in the zip have "stored" compression type; should be possible to implement with only performance loss for those few files). If anyone wants to have a shot at that I'll happily review such a patch ;) Kristian 2013/2/8 Romain Manni-Bucau <rmannibu...@gmail.com> > Hi guys, > > do you have figures regarding size and execution time? slower/bigger > doesn't speak that much to help to choose a default config ;) > > *Romain Manni-Bucau* > *Twitter: @rmannibucau <https://twitter.com/rmannibucau>* > *Blog: **http://rmannibucau.wordpress.com/*< > http://rmannibucau.wordpress.com/> > *LinkedIn: **http://fr.linkedin.com/in/rmannibucau* > *Github: https://github.com/rmannibucau* > > > > 2013/2/8 Anders Hammar <and...@hammar.net> > > > In general, I think that the default value should be whatever works in > most > > cases. Then we could have params for tweaking this (for better > performance > > e.g. in specific cases), but it would be up to the user to do this. > > So, in this specific case, I think the default should be to recompress so > > that it always works even though it might be a bit slower. > > > > /Anders > > > > > > On Thu, Feb 7, 2013 at 9:58 PM, Kristian Rosenvold < > > kristian.rosenv...@gmail.com> wrote: > > > > > A lot of you seemed to have realized that the latest version of war and > > > assembly have chosen the "fast" option over the "compact" option; and > you > > > actually seem to like it ;) > > > > > > https://jira.codehaus.org/browse/MASSEMBLY-639 has been filed and > > "fixed" > > > which will revert the behaviour back to "slow" for both war and > assembly, > > > So what do you think ? > > > > > > > > > Kristian > > > > > >
