freenode/#clim - IRC Chatlog
Search
3:27:46
nyef
beach: Here's a question for you: What is the difference between the concept for Quicklisp and the concept for ASDF-INSTALL?
3:29:38
nyef
I'm going to contend that the concept for ASDF-INSTALL is "to make it easy to install common lisp software".
3:29:56
nyef
And that the concept for Quicklisp is "to make it easy to install working common lisp software".
3:32:26
nyef
ASDF-INSTALL was popular until it broke down. Quicklisp is going to remain popular until either something better comes along, or whoever is doing the curation stops without a designated successor.
3:33:16
nyef
And it's not a matter of "just software" vs. "software and curation", but a matter of "solves an actual, painful problem".
3:36:07
nyef
Well, there's two sides to this. One is that ASDF-INSTALL completely failed to cover the actual use-case that it was trying to hit.
3:37:43
nyef
And the other is that for Quicklisp to exist, someone (Xach) had to decide that there was a problem, to realize that the solution would require maintenance over time in order to continue to solve the problem, and to resolve to *do that maintenance*.
3:39:35
nyef
Yes, deciding that a problem exists is easy enough, more or less. Correctly diagnosing the problem is harder, and that was the first failure of ASDF-INSTALL.
3:40:04
nyef
The problem wasn't "how do we install lisp stuff", it was "how do we install working, usable lisp stuff".
3:41:07
nyef
But ASDF-INSTALL quickly broke down, and now I'd be surprised if the machinery still works.
3:43:27
beach
Maybe it's more likely to find a successor to Xach than to danb because Quicklisp does solve the real problem?
3:44:17
beach
By the way, I think you missed the standing ovations when Xach was at ECLM the first time after he introduced Quicklisp.
3:45:32
nyef
It's about solving the right problem, in the right way. But it's also about *selling* the solution to both users and developers. And it's about actually doing the work.
3:46:19
nyef
Another example: SBCL. "We take the parts of CMUCL that make it a Common Lisp implementation, discard the rest, make it easier to build and maintain, and turn it into the best, most standard conformant Common Lisp implementation available for unix systems."
3:48:00
nyef
But that was basically the vision: "We do Common Lisp, starting from here, and we discard the rest."
3:49:09
nyef
SBCL is still maintained, though. stassats works on it still, dougk works on it still, I work on it on occasion, Xof at the very least still does the release management.
3:51:29
nyef
Another example: There are multiple Common Lisp LispOS projects out there. I can think of three off the top of my head, one of which is still actually being developed.
3:52:54
nyef
Some sort of vision for "this is what it means to be a LispOS" beyond "it's an OS, but written in Lisp!"
3:53:44
nyef
Because most of the LispOS discussions are basically "Let's write something that's just like ,existing-os but written in Lisp!"
3:55:15
beach
I think that's the main reason they are boring to me. I don't learn anything, and I find myself in endless debates, always about the same things.
3:55:28
nyef
It's a poorly-specified vision, for which nobody buys in on either the user side or the developer side, and then there's no execution.
3:56:43
nyef
"For a LispOS, we'll need a text editor, a web browser, a file manager, and an OS kernel. And the OS kernel is the interesting bit, so let's start with that!"