"Neil Mitchell" <[EMAIL PROTECTED]> writes: (snip) > The good and bad of each are as follows: > > 1) Move alex.exe into some global Haskell directory for binaries > (which is on the PATH) > Not too bad, but while names might not clash, if too Cabal programs > both have LICENSE in their same directory, and both read it, things go > wrong. The whole issue is that name uniqueness goes.
The .exe should get installed in a different location than something like a README, right? > 2) Make an alex.bat in some global Haskell binary directory (on the PATH) > Again, quite workable. There are a few disadvantages, the app isn't > real, so certain things don't always work exactly like they should - > for example Ctrl-C will ask "do you want to terminate the batch file" > - not just kill Alex. > > 3) Make Cabal detect wherever it puts alex, so it can then use it. > Good, and at a very minimum, Cabal should detect where it put > something. It could even keep an installed tools database, if it felt > the need. The disadvantage is you then can't invoke Alex/Haddock etc > at the command line, which is useful to have. Executables that cabal installs shouldn't just be available to cabal, though! Haskell is good for stuff besides building Haskell ;) peace, isaac _______________________________________________ cabal-devel mailing list cabal-devel@haskell.org http://www.haskell.org/mailman/listinfo/cabal-devel
