freenode/#clim - IRC Chatlog
Search
4:26:00
nyef
So, my little toy there was called "exploiter", and it killed Explorer III. Once it ran into the wall, as it were, I decided to gather more information by writing a microcode disassembler for the machine, and that became "nevermore", which was written in CL (originally CMUCL, but eventually SBCL).
4:27:10
beach
What are your current plans, if any, for some participation in a project that will move things forward?
4:27:58
beach
You were talking about the "failure of free Common Lisp implementations" the other day, for instance.
4:29:23
nyef
But I'm thinking that there's a sort of "next level" beyond having a solid base implementation (SBCL, CCL, whatever), and a curated software installation system (Quicklisp).
4:31:39
nyef
So I'm simultaneously trying to figure out an overall metastrategy, this vision-sales-execution thing, and trying to figure out what using CL should be like *as a concrete experience*.
4:32:20
nyef
One of the provocative questions that I'm using is "what do I not do because even the thought of making the attempt is too painful to contemplate?"
4:33:58
beach
But it is usually when trying to debug some problem, and I find that my tools are not up to the task.
4:34:50
nyef
Here's one main thing: I don't do UI work, because none of the toolkits are particularly usable.
4:37:07
nyef
Part of the problem is that many of them are FFI based against a moving target of a library.
4:38:35
beach
And they are related as you pointed out: "just look at the C documentation and replace underscores by dashes".
4:38:46
nyef
Having used my MacIvory for less than an hour, I'm already wanting to hit Select-D on my Linux box.
4:40:33
nyef
I don't know that it's entirely appropriate for a modern environment, but at least it's there.
4:42:12
nyef
Every so often, we find out about some lone hacker or small team that has put together a phenomenal amount of software, sometimes with documentation, but nobody else uses it but the originators.
4:43:08
nyef
Probably at least partly because it's stacked several dependencies deep, and has its own worldview which isn't exactly compatible with how anybody else does things.
4:44:58
nyef
Right, and that's fair. But it means that if he *does* start writing stuff that you're interested in, it's probably on top of this huge pile of weird.
4:45:56
beach
But I am willing and able to read documentation, as long as it doesn't mean I have to understand some C++ code or similar.
4:47:45
nyef
Basically, it had a fairly substantial chunk of reference material available as hypertext, with example programs.
4:48:32
nyef
These were the days where, if you were selling a compiler, it was a *product*, and it had *documentation*.
4:51:00
nyef
Having the same environment, as near as possible, over *all* the ways of running your product.
4:52:13
nyef
If it's in a GUI system, it's a separate window with whatever. But in a terminal window environment, it still behaves the same, same input gestures (except possibly for the mouse), same editing keystrokes, and so on.
4:54:15
nyef
So, I think that what I want is that level of "product" thinking applied to a Lisp system.
4:56:34
nyef
Which means that, aside from a base implementation and quicklisp, it needs a UI framework, some sort of story for documentation (possibly both a lisp-side documentation browser and something to export to a web browser), an IDE (both code editor and debugging tools), and some *good* technical writing.
4:57:29
nyef
Which is enough work to keep a team of programmers (and technical writers) busy for some time.
5:00:18
nyef
We even sortof need *three* UI stories. One for a "lisp-like" UI, one for a "normal, but possibly ugly" UI, and one for "just like a native application on your host machine".
5:04:29
nyef
Well, yes-and-no. The "lisp-like" UI is CLIM, exemplified by the classic Listener window. The "normal, but possibly ugly" UI is still CLIM, but configured to look more like a standard application, rather than something with a command prompt.
5:05:11
nyef
And the "just like a native application" possibly involves some of the CLIM theming machinery, or even straight-up switching backends.
5:06:10
nyef
Part of the problem with #3 is that "just like a native application" often means interfacing with an FFI-based toolkit.
5:08:01
nyef
And, worse, OpenGL is an important use-case, so we may need to either find a way to use host OpenGL libraries with a Lisp CLX without needing two X server connections, or we may need to use xlib / xcb via FFI.
5:09:54
nyef
There's also the bit where macOS and Windows don't use X11, and there's even efforts to move Linux away from it as well.
5:12:57
beach
I have also been thinking about a kind of intermediate library for drawing things so that one could use either X11 or something else behind it. But this is not really my domain, so I often don't get very far.
5:15:32
nyef
It's just coming up on quarter past one in the morning here, so I'll probably be off soon.
5:18:13
nyef
Basically, though, while I know that I want something in terms of a better Lisp experience, and I have a vague notion of what that something might be, I don't have a specific enough vision that I can plan it out, sell it (to myself first of all), and see to arranging for execution.
5:21:52
nyef
I think that this is a good start, though. I'm going to have to copy this into a document for further editing.
9:57:44
MonsterHesh
shka_ vydd smokeink oleo scymtym Reinisch phoe DougNYC terpri brucem Faed red-dot paroneayea slyrus payphone jackdaniel jdz mood fiddlerwoaroof ecraven loke``` nyef groovy3shoes nicdev TeMPOraL mood__ PuercoPop eschatologist Colleen jurov gko p_l knobo splittist ChanServ christiansen`