The biggest issue with marshalling is that we simply don't do it right...and there's NO SPEC for how it's supposed to work in Ruby. I almost launched into writing one by reverse-engineering the C code several times, but other things took priority. I think it would tremendously helpful if someone could write or find a binary spec for Ruby marshalling. The implementation would be trivial if we had that.
What I found when I started trying to fix things was that there's just a lot of pieces we don't even have code for. The FlashHash < Hash thing was just the beginning. The complication is that we have to figure out how it's supposed to be done solely by looking at the C code.
The line terminator thing just sucks like the path-separator thing does. We'll have to figure out where those inputs and outputs are and do the conversion as necessary. I wish there was a flag we could set in Java to just run it in "unix mode", but that would probably break anything other than JRuby running in the same JVM. *sigh*
On 6/19/06, Thomas E Enebo <[EMAIL PROTECTED]> wrote:
On Mon, 19 Jun 2006, Ola Bini defenestrated me:
>
> Ah, yes, I know where you're coming from here.
> I have some reservations about the line separator issue too, but nothing
> serious enough to keep the patch out for now. We'll have to fix the issue
> correctly at some point, though.
Yeah that is mostly how I feel at this point.
> Marshalling is somewhat tricky, but when I last checked the code, it didn't
> seem to be that much of a problem. What are the complications? Is the error
> with FlashHash that our marshalling doesn't dump the class name for native
> classes?
class FlashHash < Hash
attr_accessor :something
end
With trunk this ends up just saving a Hash. When it reads it in things
are unhappy since it wants a FlashHash. Charlie had a patch which
saved the right class name, but it did not do a couple of other needed
things. I have something which right now remakes the FlashHash but an
extra instance var (which is a Hash) is getting set to nil. I am almost
there. I am hopeful this will not cause problems elsewhere since it will
really only trigger with user-derived hash classes.
> But if there is something really important, throw me a mail and I'll see if
> I can find some time.
Will do if needed.
> Of course, it feels great with all the current progress!
>
> (Did you check the new Signal version, by the way?)
I did not look with any amount of detail yet, but I plan on committing
this first thing when 0.9.0 opens up.
-Tom
--
+ http://www.tc.umn.edu/~enebo +---- mailto:[EMAIL PROTECTED] ----+
| Thomas E Enebo, Protagonist | "Luck favors the prepared |
| | mind." -Louis Pasteur |
_______________________________________________
Jruby-devel mailing list
Jruby-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jruby-devel
--
Charles Oliver Nutter @ headius.blogspot.com
JRuby Developer @ jruby.sourceforge.net
Application Architect @ www.ventera.com
_______________________________________________ Jruby-devel mailing list Jruby-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/jruby-devel
