Recent blog entries for sublum

well, terminals aren't fun, but they work.

installed QNX recently. Its Photon GUI looks nice, but the pkg-installer keeps crashing on me; this makes it hard to go anywhere with it. Some more "issues", minor or not:

• the dialer is too limited, on (1) what it allows for a modem init string and (2) what it reports of responses from the modem peer
• the Shelf configurator doesn't (yet) allow you to "move" components within the shelf hierarchy
• the package system could really use some influence from Debian's apt/dpkg setup; it's really a hassle in comparison to using apt/dpkg/aptitude; the QNX pkg-installer doesn't seem to cache the downloaded repository-info files (yet), across sessions, and it seems like it'll be far more work than should be necessary, to dig-up enough repositories from the web, for getting all of the dependended-on packages, which are needed for software in the already-listed package repositories. GTK? xlib? where? (this, on top of the frequent crashes of pkg-installer, make it hard to build a useful installation with it)

Otherwise, I've heard that their development-kits are great; looking forward to when I can, if ever, fit them in with some open-source tools -- to start with, a Common LISP implementation, either CMUCL or SBCL, if i can figure out how to get one or the other bootstraped on an OS where there might not already be an available package of one or the other.

admin-space: clean, no fluff, no extraneous apps, direct, and to the point

user-space: hold their hands and baby them??

straddling "user space" and "admin space", working to make a [Linux? BSD?] distro that works clean and perfect in both ways/spaces .. release-mangement, to start with .. need something for tracking projects and "requirements specifications" and proceeding from the latter to a finished product .. need tests, based on the requirments specs, beng run at every branch (and even every checkin) along the way

need to learn how to write an LDAP schema first?? what the heck is that about?

And then, there's QNX, but the set of Linux and BSD seems like a heck of a lot more open of a market, in "business terms".

"bill gates didnt do it right, he just made a ton of money making people think he did"

-- daos, irc.openprojects.net/#labyrinth

is there no such thing as a cyclic order?

"does there exist an ordering that results in a cyclic relation among the ordered elements?"

someone can tell me "yes", in answer to the first question, and "no", in regards to the second, until someone runs hoarse from repition of the same, but it's not likely that i'm going to believe that someone -- now, and probably later.

it would be completely foolish of me, if my reasoning for that is based on hothing but "superstition", or if it's based on nothing but a tendency towards "dislike", of who might try to convince me otherwise [ad hominem], or a combination of those and any other useless motivations, but i cannot yet present a concrete proof for the supposition that "there exists an ordering for a set S, such that some [perhaps all] elements of set S exist in one or more cyclic relations to each other", and yet i am not willing to dismiss it.

this isn't the right place to write a rough draft in any attempt to better address this.

the snowball effect, in the key of D minor:

"1001 incomplete projects on the wall, 1001 incomplete projects; take one down, start working on it, 1010 incomplete projects on the wall"

Revision Control Systems

• CVS

  CVS is just barely useful for web-site development. It makes it a royal pain in the foot, to move a directory and all of the files in it.

• Aegis

  Aegis is great, except for a few major points:

  • The removal of the working-directory, on check-in
  • Some often-not-needed points of the Aegis author's concept of how revision-control should happen.
    • the enforced "reviewer" step
    • the enforced "integrator" step

    ...both of which are a pain in the foot, when the developers are fit for doing their own review and integration... which they sure as hell should be! in any case.

• Subversion
  1. that was a really unfortunate choice, for the name of it.
  2. still in the pre-alpha stages of development

stay away from Epson printers.

the ink cartridges on Epson Stylus C60, C80, and 820 models are non-refillable, "because of the new micro chip."

[reference: printerfillingstation.com]

just figured out something that i really like about Lisp, in direct comparison to Java(tm) (and, for that matter, C++ and some other languages):

making forms of a 'foo' function or method that would work for both (foo a) and (foo a b) is as simple as [pseudocode]:

(defthing foo (first &optional second) DEFINITION)

...and defining it all in DEFINITION.

this is considerably cleaner than 'overloading' foo, as so:

etc etc foo(first) { definition }

etc etc foo(first, second) { definition }

and that's part of (1) why I use Lisp, and (2) why Java(tm) drives me mad.

imagine a window-manager, with a theme-manager|theme-editor app that generates ASM code, and uses NASM (INCBIN) to include any images and other data, for a theme, into a compiled library that would be the 'theme' itself.

Besides avoiding some disk-read calls, that, by itself, does not necessarily seem like it would be any more effective than storing image files on disk, and reading them in to memory when the wm loads up. However, there are some usage cases, in which it would better to "do that library thing" with the theme files. For instance:

• an "embedded" window-manager, on a device that effectively has no disk-type storage medium.
• anything "embedded", if it could use some user-specified (as opposed to "hard-coded") graphical user interface settings
• cases on "non-embedded" type systems, where the number of distinct theme images is very large
• cases -- not necessarily with NASM -- where some sort of "code obfuscation" techniques might be seen as necessary, for ~encrypting the original theme data

some things are on the board.

• CLEM
• ASM
• 3Dwm
• business-plans in DocBook xml?
• trying to set up a net to catch some employee redux, if AOL buys RedHat? and finally putting "a commercial layer" on Debian, and feeding some money into the Debian project, and to the package-maintainers?
• sometime, eventually, maybe, selling the combination of Linux and Common Lisp?????
• "hey, why don't we [i] start (inching our [my] way towards) working on an embeddable Lisp implementation, ior a full, free LispOS?"

??