headers
Attila Lendvai | 7 Nov 2009 13:15
Picon
Gravatar

Re: Patch to rfc2388 to make IE file uploads work properly

>> can anyone else confirm that our rfc2388 is wrong and ie is right?
>
> any word on this?
>
> this still cleanly applies to the rfc2388-binary repo...

fyi, i've pushed these into the official repo and we'll see... still a
much better place to rest than the mailing list.

--

-- 
 attila
Permalink | Reply | 0 comments |
headers
Attila Lendvai | 7 Nov 2009 13:28
Picon
Gravatar

Re: [PATCH] Allow use of macrolet in with-call/cc compiled to FASL on SBCL

> As subject, this is an improved version of the patch I sent a few
> months ago.
>
> At the moment SBCL barfs, complaining it cannot serialise a function
> into a fasl if you use macrolet inside a with-call/cc and compile to a
> fasl.
>
> I don't know whether it is sensible to write a 5am test for this as
> you need to compile something to a fasl.

fyi, we've cleaned up and factored out the walker and the call/cc
stuff in arnesi. they are now in hu.dwim.walker and hu.dwim.delico
available at http://dwim.hu

in this codebase the walker does not store the macro functions in the
bindings of the macrolet, because it doesn't make much sense. macros
are macroexpanded at walk time, so the definitions only cause trouble
while not really useful for anything.

i'm even considering walking macrolets unconditionally into locally
blocks, but that may hinder debugging a bit.

anyways, thanks for the patch!

--

-- 
 attila
Permalink | Reply | 0 comments |
headers
Attila Lendvai | 9 Nov 2009 11:00
Picon
Gravatar

Re: [PATCH] Allow use of macrolet in with-call/cc compiled to FASL on SBCL

> I moved to cl-cont because of too many Arnesi bugs, but I guess I should
> give it another try now.

well, there were not that many call/cc changes, but we made several
fixes/improvements to the walker.

my (not imminent) plans with delico is to add closure based
continuations like cl-cont besides the interpreted (and serializable)
one which is currently provided.

> Do you run something like the ANSI test-suite to verify your walker is
> correct? (Before going further with macroexpand-dammit/cl-cont I want to
> test it thoroughly.)

unfortunately not. it would be a good addition to the delico tests.
i've checked out the test-suite repo, but it's not on the top of my
TODO...

--

-- 
 attila
Permalink | Reply | 0 comments |

Gmane