Re: [Swarm-Support] method `dropAllocations:' not implemented by protocol ?

[amores nolikeyjunk]

> I don't know what you mean, "all binary versions are broken."

I'm sorry; I just can't make anything work. I may be doing
some things wrong, but it appears to me that there
are problems in the tarball itself (I fixed one problem
by copying code from the cvs version back into the pretest
tarball, but this makes me wonder if it works out of
the box for anyone ? Altho, I wonder if these ought to
just be warnings, and I have a bad gcc -- that would make
more sense I think.)

> Nobody is maintaining Windows at the moment, that's all.  I tried to build 
> Swarm from scratch on Cygwin and gave up in disgust a summer-and-a-half 
> ago, but Marcus Daniels, who is much better at this stuff than I am, did 
> succeed, and he was the one who was packaging up the pre-tests of Swarm.

Yes, you may no longer recall, but you had sent me email a while
back (when I first set out hopefully trying to install swarm)
telling me that there was no one maintaining the cygwin version.
I kind of felt encouraged to try to make it work, but I may have
misunderstood; you probably weren't encouraging me, but only
stating the state of affairs. In any case, I think it does prove
to have been quite naive for me to expect to succeed where you
have tried so hard & failed. I had naively thought, ok, well,
maybe it didn't work for him a year ago, but now cygwin and all
the tools will have magically been fixed, and everything will
turn out splendidly now. Then when that didn't prove to be
the case :(, I kind of kept going, thinking next, perhaps if
I get and build the latest gcc, that will prove to make
everything work. I had no idea there were problems with gcc and
Objective C; as I don't know Objective C, it is probably
laughably futile for me to be pursuing the attempt to compile.


> If somebody has a working Windows system with Swarm-2.2 and  patience, I 
> believe what they need to do is take a Swarm tarball and rebuild with a 
> configure command so that the prefix so it installs into the same place 
> where Swarm-2.2 is.  Then tar up that directory, and a new Swarm pretest is 
> born.

Here is, I think, a place where I'm misunderstanding. I think the
so-called "stable" version on the mainpage is Swarm-2.1, and
I haven't tried it (because I thought the mainpage said it is
advised not to use it, so I feared that even if I made it work,
as soon as I had some question, I'd find the natural response of
"well, first you should follow the directions and get the
development version", and then I'd have to unlearn out-of-date
stuff, and relearn from the development version--therefore it
seemed clever to me to skip that whole step & jump ahead to the
development version :))

Is the Swarm-2.2 version then the binary distribution ?
The one that doesn't work with modern cygwin because modern
cygwin doesn't have cygtcl80.dll ? If so, that was the first
thing I tried. I only went to source compilation because
that didn't work, and I couldn't find source to try compiling
Tk/Tcl myself (altho I didn't try too hard, as I misunderstood
you to have encouraged me to try to compile from source instead;
I'm pretty sure this was entirely my misunderstanding!). I did
later, after you kindly redirected me to this mailing list,
see a link where you made available an entire Tck80 cygwin
package, and I excitedly tried it, only to find segfaults
immediately with everything I tried (well, only two samples;
I was immediately discouraged). I didn't try to debug them; I
don't think I'd be competent debugging either Tcl problems
(not knowing Tcl) or Swarm problems (knowing neither Swarm nor
Objective-C). Quite possibly I missed some needed step in
bringing the Tcl80 files into my existing cygwin tree.

I did get a RedHat version up, but it won't help me for very
long even to have a working Swarm on RedHat, because I will only
be able to use Swarm if I can make it work for others on Windows.
I had thought to use the RedHat system to try to work out what
problems are really in the tarball (the two different missing
methods errors I've seen -- I'm not sure if those are bugs that
need fixing in that distribution of Swarm, or if rather I
just need a different set of flags for gcc to not flag them
as errors -- the latter would make more sense, as I'm guessing
that tarball works for other people).

I'm going to try using the obsolete binary distribution. I had
earlier thought it would fail for me the same reason the
development binary version failed (because modern cygwin lacks
cygtcl80), but having looked at it again, I think it might
include all of its own cygwin, so it might work out-of-the-box
for me. As such a newbie, that might be my best course for the
present.


> Now, about Windows, I've got a WinNT system that works with the older 
> tcl/tk, but sometime I may try to build a new Swarm pretest snap on there.  
> But I don't have much of a reason to do it. I'm just a volunteer here, not 
> an employee.  My attitude is: I switched over to Linux in order to use 
> Swarm, so why should I have to suffer for Windows compatability?  If 
> somebody else would step forward and volunteer to maintain a  WIndows 
> version of the Swarm FAQ, I would gladly forget MS ever existed.

Yes. I thought quite naively maybe I could make something work,
but I hadn't realized how many different issues are involved (it
sounds as though there is quite a problem just finding a version
of gcc that will work in the first place), and I don't think I'm
qualified at the moment, unfortunately.

But, thank you for all the help, nonetheless.


Cordially,

Perry






_________________________________________________________________
STOP MORE SPAM with the new MSN 8 and get 2 months FREE*  
http://join.msn.com/?page=features/junkmail