[Openmcl-devel] save-application and mcclim

[James Ashley]

I was missing something obvious.

I had my environment variables set up point at the directory where I
extracted ccl. I copied the ccl script to an executable directory
where I could run it. I was saving my images, and copying them into
that directory next to the kernel. And the kernel was happily loading
the image from the path where I extracted ccl in the first place.

*blush*.

When I placed the image file (with the proper name) inside the
appropriate directory where I'd extracted ccl (and set the environment
variable to point to), I could try to execute (climacs:climacs) and
get a "socket closed" error, which completely makes sense.

Yes, I feel like an idiot for introducing myself this way.

Thanks to anyone who listened to me babble (it's been a frustrating evening),
James


On Sat, Nov 13, 2010 at 6:50 PM, James Ashley <james.ashley at gmail.com> wrote:
> Oops, that probably wasn't a good test-case (as far as the file sizes
> go). I have things set up to load quicklisp automatically on startup
> now. Which is probably muddying the waters, but may also express
> another symptom.
>
> Get rid of the .init that loads quicklisp automatically. Do
> (defparameter *check* 9) and (save-application "whatever") again.
>
> The image file is 26104048 bytes this time. Same error on *check*.
>
>
> ---------- Forwarded message ----------
> From: James Ashley <james.ashley at gmail.com>
> Date: Sat, Nov 13, 2010 at 6:11 PM
> Subject: save-application and mcclim
> To: openmcl-devel at clozure.com
>
>
> Howdy, everyone. n00b warning.
>
> This could very well be the totally wrong list for this topic. Or
> maybe it's an FAQ that I missed. Or maybe google did a horrible job
> digging back through the archives. If that's the case, I apologize.
> Somehow, I don't think this is the case.
>
> I'm trying to get a ccl environment up and running on linux (yes, I
> know). I got the kernel and image running (seems to be working
> fine...emacs and slime seem to be perfectly happy, even if several of
> the examples have issues...I figure one of the best things I can do to
> learn/contribute is to figure out what those issues are and how to fix
> them).
>
> Then I installed quicklisp (I know, beta, 3rd-party, unsupported
> software, bear with me), set it up to load automatically, and used it
> to install climacs (*sigh*...yes, I *know*).
>
> Up to this point, everything's working great.
>
> climacs takes forever to load, so I did a (save-application
> "whatever"). Then replaced the image file in the directory with my
> kernel with that file.  Restart...no climacs.
>
> So, I restarted with a basic image.
>
> (defparameter *check* 9)
> (save-application "whatever")
>
> Restart emacs and slime.
>
> *check* -> "Unbound variable *check*\n\t[Condition of type UNBOUND-VARIABLE]"
>
> (yes, I've spent way too long in the c++ world).
>
> The images that are being created are significantly bigger than
> default supplied ones. 66666736 bytes for the "with clim" one and
> 28303600 bytes for the "with a dynamic variable" one, as opposed to
> 735459 for the original. Not that I care about that...it just seems
> like it might be a symptom.
>
> Should I switch to the 1.6 pre-release version? (That's sort of my
> inclination, but I'm already feeling on the lonely side of the
> bleeding edge). Am I just missing something really basic? Is there
> something I could do that might offer someone else a clue about where
> the problem lies (especially if it's PEBKAC) that might point me in
> the correct direction? Should I switch back to SBCL until I'm not so
> clueless? Break down and switch to Windows (uggh)? (Right now, a Mac
> just isn't in the budget).
>
> Thanks in advance,
> James
>