[prev in list] [next in list] [prev in thread] [next in thread] 

List:       freebsd-java
Subject:    Re: Plugin Status
From:       Bill Huey <billh () gnuppy ! monkey ! org>
Date:       2001-11-30 22:18:37
[Download RAW message or body]

On Fri, Nov 30, 2001 at 10:22:54AM -0800, Joe Kelsey wrote:
> I am supremely tired of fighting with Motif bugs.  The plugin currently
> loads, reads and responds to commands from Mozilla, and then dies as it
> tries to create a window within Mozilla's window.
> 
> I have tried it with both Open Motif and LessTif.  LessTif is much
> better, since it has better diagnostics by default and fewer blatant
> bugs.  Open Motif is such a hog--poorly written and littered with
> blatant programming errors.  Sigh.

Did you try and talk to the OpenMotif folks ? They're pretty helpful and
are willing to fix bugs quickly from my experiences.

> I plan to put aside the plugin for the time being and switch over to
> working on HotSpot.  The last time I switched gears, from trying to
> debug green threads to using Bill's native threads, the plugin operation
> suddenly improved greatly.  I am hoping that something to do with
> HotSpot may have a similar effect on the plugin operation.

Hehe, that's good to hear. Geez, I hope folks realize how significant
this is for the overall port. It basically means we are "for real" now,
IMO. ;-)

> HotSpot is organized into system-specific directories.  The current
> split is between solaris and linux, with most files duplicated in the
> separate hirearchies.  After seeing some of the horrible linuxisms
> contained in the current code base, I am going to go and base the bsd
> code base for HotSpot on solaris and move forward from there.  Every
> time I look at something done for Linux, I get sick.  I may refer to the
> linux code if I need a laugh, but otherwise I plan to more or less
> ignore it.

Yes, it is. It's a pretty standard about machine specific conventions
like with all JIT compilers. I'd be worried about JIT block linkage and
calling conventions within HotSpot itself since GCC is going through
significant changes and it's unclear how the -stable compiler will behave
with HotSpot. It's a pretty self contained system, but having it written
in C++ is potentially a serious problem considering that GCC is constantly 
changing core C++ internals every minor revision or so it seems.

Good Luck and report back errors either publically or privately. ;-)

> For now, I am recompiling LessTif with debugging enabled in one last
> effort at finding bugs.  Then it is on to HotSpot.

I recommend bailing on LessTif and work with the OpenMotif folks. Post
a stack trace to them and they should be able to give you some solid
pointers on the subject and possibly build a prepatch for you that fixes
the problem. Tell them who you are and what your relationship is to this
project, how significant it is for the BSD community and I'm sure somebody
will get back to you on the subject. Definitely try and people network
with people as much as you can when doing ports like this which are too
large for a single person to fully understand. That's when talking to
other folks really facilitates the development process.

[Personal Topics]

I might be taking a job soon as a WINE engineer. If that's the case, then
I'll be dropping out of this development group and until my next development
cycle.

I'll keep you folks informed. My role here is pretty close to coming to
a point of completion since native threading is largely working.

> /Joe

bill


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-java" in the body of the message


[prev in list] [next in list] [prev in thread] [next in thread]