Login | Register
My pages Projects Community openCollabNet

Discussions > dev > Re: [helm-dev] compiler default

Project highlights: Architectural Overview

joist
Discussion topic

Back to topic list

Re: [helm-dev] compiler default

Author edk
Full name Ed Korthof
Date 2000-12-15 01:00:16 PST
Message On Thu, Dec 14, 2000 at 10:43:09PM -0800, Jon Stevens wrote:
> on 12/14/2000 3:49 PM, "Manoj Kasichainula" <manoj at collab dot net> wrote:
>
> > On Thu, Dec 14, 2000 at 02:19:55PM -0800, Ed Korthof wrote:
> >> i'd like to change the build-*.sh scripts so that jikes is the default
> >> (if it can be found) rather than javac.
> >
> >> of course -- checking to see if jikes is installed is a bit of a pain --
> >
> >> the logic is : if JAVAC is set, use that (whatever it is). if not, and
> >> if JIKES is set, use jikes. if neither of these is true, use which to
> >> look for jikes in the PATH; if we find it, use jikes. otherwise, use
> >> javac.
> >
> > This sounds like a lot of work...
> >
> > How about just running $SANDBOX/bin/javac, and have that be a symlink
> > or script pointing to the appropriate compiler?
> >
> >
> How is that more work than just setting an environment variable?
>
> Actually, I would simplify the above by simply doing the following:
>
> look for JIKES=yes
> if JIKES == yes
> use jikes
> else
> look for JAVA_HOME
> if -z JAVA_HOME
> then fail
> else
> use JAVA_HOME/bin/javac

looking at this -- i find it really weird that you set
'JIKES={anything}' in order to set JAVAC to jikes. why not simply set
JAVAC directly? i started with a feeling of unease wrt that.

besides that -- the fact that the shell scripts default to javac is
annoying in a development environment. sure, there are ways to fix this
-- but it's been a pain on our shared dev boxes. with sc-dev1, i set
things up so that bash users would default to jikes ... we can continue
down that path, doing an extra step in system install/configuration for
each new box to solve this, or integrating this step into our rpms
(thereby poluting the environment even more than we already are).

but when i start imagining that -- i have to wonder if it makes any
sense at all. why would *anyone* want to default to javac if jikes is
available? if there's no good reason to do so, then defaulting to jikes
(when its available) makes sense. i don't care if we use the approach i
outlined or another; but i see no reason not to try to make life easier
for our users.

moving the complexity to system install or individual user configuration
doesn't remove it -- it just shifts the burden. imo, in this case, it
produces a net increase in the work required (unless JIKES=yes is a
standard way in Java softare to choose jikes over javac).

cheers --

ed
--
   +=-=+=-=+=-=+=-=+=-=​+=-=+=-=+=-=+=-=+=-=​+=-=+=-=+=-=+=-=
   | Ed Korthof | edk at collab dot net | 650-228-2527 |
   +=-=+=-=+=-=+=-=+=-=​+=-=+=-=+=-=+=-=+=-=​+=-=+=-=+=-=+=-=

« Previous message in topic | 8 of 8 | Next message in topic »

Messages

Show all messages in topic

compiler default edk Ed Korthof 2000-12-14 14:19:55 PST
     Re: compiler default edk Ed Korthof 2000-12-14 14:24:15 PST
         Re: [helm-dev] Re: compiler default dlr Daniel Rall 2000-12-14 14:48:50 PST
     Re: [helm-dev] compiler default Manoj Kasichainula <manoj at collab dot net> Manoj Kasichainula <manoj at collab dot net> 2000-12-14 15:49:24 PST
         Re: [joist-dev] Re: [helm-dev] compiler default Jon Stevens <jon at latchkey dot com> Jon Stevens <jon at latchkey dot com> 2000-12-14 22:43:09 PST
             Re: [joist-dev] Re: [helm-dev] compiler default deploy Deployment Pseudo-user 2000-12-14 22:48:49 PST
                 Re: [joist-dev] Re: [helm-dev] compiler default Jon Stevens <jon at latchkey dot com> Jon Stevens <jon at latchkey dot com> 2000-12-14 22:59:02 PST
             Re: [helm-dev] compiler default edk Ed Korthof 2000-12-15 01:00:16 PST
Messages per page: