> fwiw, I have found UPX to be a piece of crap for most applications. Their > site shows that the idea was conceived and implemented during the days of > Pentium I computers, as all their CPU tests show that sort of historical > period. > > You'd think that would be a good thing, but at the time 16 MB of ram a pretty > standard amount to have... which leads me to believe that in order to avoid > using RAM for UPX exes, they might have been uncompressed via the filesystem. > They boast hardly any memory footprint, so I believe something tricky is > being done. > > Problem is, on computers 10-15x that we have around the office, which have a > clock speed of probably 10 to 15 times that, the UPX exe's over a 100mbit > network sometimes take WAY longer to load... in the seconds, compared to > almost a fingersnap. I'm not sure if my wild guess above is the cause for > it, but I tested this on about 5 computers at work to decide to stay away > from UPX, despite its nice reduction on our .EXE sizes... > > I know this probably isn't relevant, but I thought I'd mention it anyhow, in > case someone else has noticed this.
That's interesting experience because the only reason I still use UPX is to shorten load times on networks, based on the theory that serving 2.5MB from the file server is quicker than serving 14MB. Never made any load measurement though on a network. The only noticeable effect on local machine is somewhat slower loading time due to lzma decompress phase. [ I use '-9 --lzma' mode. ] [ The other good side effect is distribution size is smaller when using UPXed .exe since it uses lzma while for distro I still use .zip. The reason for .zip is that it's the best we have built in Harbour, and also the LCD. ] Anyhow now I'm distributing without UPX and half of me is happy because dropping a tool from the toolchain is always nice, plus I find it more "honest" to distro vanilla executables, antivirus kits like it better, too. So no huge loss, but it's good to know. I'd glady hear about more UPX experience, anyone? Viktor _______________________________________________ Harbour mailing list (attachment size limit: 40KB) Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour
