Re: cl-ironclad
Hello,
On Fri 26 Jul 2024 at 08:20pm +03, Christoph Berg wrote:
> Re: Sean Whitton
>> cl-ironclad very often breaks when I upload sbcl. But it doesn't really
>> have a maintainer. I would like to remove it, in the absence of someone
>> willing to work on it.
>>
>> Unfortunately this would break pgloader.
>> Do you think pgloader is still used much?
>
> Hmm. It's still being used, but not very widely. It wouldn't be the
> end of the world, but still sad after all the work we put into it and
> the lisp ecosystem in Debian to make it work.
>
> I was just looking through the list of packages build-depending on
> sbcl and found only two actual applications - stumpwm and pgloader.
> Removing cl-ironclad would kill 50% of the sbcl users in Debian. We
> could do it, but wouldn't that basically admit that lisp isn't fit for
> actual application development? (Or does it just say that ironclad is
> the wrong library for crypto?)
My Consfigurator is another SBCL thing that's in use :) And people have
their own projects.
Anyway, I think the problem is ironclad, not Lisp. If you look at the
README there are a bunch of warnings saying "not always safe". Indeed,
for crypto with Lisp ISTM that we should be relying on CL's strong FFI.
>
> Re: Sean Whitton
>> ... or if we can't remove it overall, can we remove it on 32-bit archs,
>> especially i386?
>
> That sounds much better. I haven't yet decided to kill all 32-bit
> support for PostgreSQL in Debian since it actually does work fine, but
> server extensions are becoming more and more a problem since people
> don't test their code on non-64-bit (or non-little-endian) platforms.
>
> So yes, that would be fine.
Okay cool, thanks, we'll do that.
--
Sean Whitton
Reply to: