input line is too long

Dhek Bhun Kho bhun@chello.nl
Mon Dec 30 16:35:00 GMT 2002


Hi Tom!!! 

Op di 31-12-2002, om 00:08 schreef Tom Tromey:
> I think I compared jikes and gcj -C a year or so ago.
> As I recall they were comparable in performance.
> This was nothing scientific, just compiling some random library with
> both compilers and `time'.
> Tom
Best wishes to you and everyone the list, I finally go my system up and
running (not totally from scratch, but hey it got pretty near to it,
waiting like 2 days to get everything compiled the way I want it to) :D

As the subject is still "input line is too long" I thought I'd better
answer it. I couldn't make much make sense out of the GNU make manual (I
get the point reading it, but how to get it into practice?? Getting the
-MM -MF output into a make file is really freaking me out.) I just run
it with 

$ go <srcdir> <version>

I rewrote my bash script to be more general so you can unleash it on any
source tree, it uses some regex and some simple comparison to prevent
regenerating every single class file<erik?>, but it's kind of slow
because it uses find to look for files, should work on any *nix platform
with bash2 and find though. It just serves my basic purpose, it's not
really adequate for heavy duty use I guess. 

The strange thing is that:
1) when generating object or class files from java sources, less errors
occur
2) doesn't GCJ just need the classfiles to find the function prototypes
when generating native code? or am I missing the point? why doesn't it
suffice to just include the CNI headers?

Best wishes.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: go
Type: text/x-sh
Size: 7131 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/java/attachments/20021230/039dbd88/attachment.bin>


More information about the Java mailing list