tychoish - Latest Comments

Re: On Wanting a Kindle

aetataureate — Thu, 12 Nov 2009 09:50:27 -0000

P.S. Related thought that occurred to me after posting.

I very, very rarely pay retail for a book since I buy most of them used. What if Amazon opened some kind of marketplace for books you'd read, so you'd get part of your money back and the buyer would get a discount? THAT would be ideal.

Re: On Wanting a Kindle

aetataureate — Thu, 12 Nov 2009 09:40:22 -0000

This post makes you sound like a fortyish woman desperate for a baby. "I'm not having one today, but I really REALLY REALLY WANT ONE. I've held them before and they just feel so nice."

As far as your concerns, I share them. The books are overpriced and the DRM is scary (Amazon can remove books from your Kindle without notice OR a refund? Really?). In my estimation the best reason for me to potentially get a Kindle is to have a daily newspaper delivered into it. And is that worth $300 . . . ?

Re: Interview with Michael Pobega

madalu — Wed, 11 Nov 2009 12:30:14 -0000

Sorry to be a contrarian, but as someone who is proficient in both vi and emacs (I was a full time vim user for three years before switching to emacs last year), I find emacs just as efficient for editing text with a few lines.

I often hear the claim that vi is faster editing code from vi users who have never used emacs for any length of time. But this seems an impressionistic judgment based on personal preference and familiarity. I would be curious to have some hard evidence about how vi is superior in this regard.

Re: "integrated regexps": Emacs certainly has this (query-replace-regexp), though its regexp syntax is admittedly a bit eccentric. But then Emacs has the ever helpful re-builder for testing out regexps on the fly. And, in the end, it's all those wonderful extensions that make emacs so powerful.

Oh, and then there's the community. Just compare these two wikis:

http://vim.wikia.com/wiki/Vim_Tips_Wiki

http://www.emacswiki.org/

;)

Re: Interview with Michael Pobega

tychoish — Wed, 11 Nov 2009 12:02:17 -0000

In these situations, my summary of the emacs/vi debate is as follows:

They're really editors that are attempting to accomplish very different
sorts of tasks, so while they're both "text editors" and you can
accomplish very similar sorts of tasks in both. However, I think it's
unwise to really compare them nose-to-nose, given their differing
gaols.

For quick editing, vi can't be beat. And not just because vi loads
faster than a fully configured emacs, but because the navigation and
normal/insert mode differences are really designed to make editing a
few lines in a file much easier. Integrated regexes (the sed style
find and replaces, etc.) also further this. And it's not that emacs
can't do any of these things, but I think the vi approach faster and
more efficient from a user perspective.

Emacs, by contrast, makes editing text, and not just any text, but the
kind of text that I deal in so much easier. It supports huge files,
the "single mode+modeline" interface makes it easy to just edit files
organically while still preserving a great deal of interactivity with
the application itself.

Another commenter here describes emacs as "the last of the lisp
machines," and while I don't think you can really convince someone who
isn't sold on the whole "lisp-thing" that programing a text editor in
lisp is a good idea, it totally is. :-o

Cheers,
sam

Re: Interview with Michael Pobega

madalu — Wed, 11 Nov 2009 09:32:44 -0000

Thanks Michael. Good stuff.

Please pardon the following flamebait, but I couldn't resist:

The allegation that emacs lacks a good text editor is something that vi users like to tell themselves as compensation for the fact that vi lacks its own programming language. ;)

I use both vi and emacs and love them both. But I find emacs every bit vi's equal as an editor -- with the benefit that it is endlessly extensible.

Re: English Songsters

Polycarp — Fri, 06 Nov 2009 18:23:51 -0000

If you want to hear "A Bottle of the Best" you can click this link:

http://www.stickam.com/viewMedia.do?mId=182240874

It might be the greatest drinking song of all time and, like you I think one of the best things about the song is that it makes fun of not only what the English and Irish drink, but their drinking customs in general.

Re: Radicalism in Free Software, Open Source

madalu — Wed, 04 Nov 2009 15:03:45 -0000

Very well said!

I've been distressed by the amount of RMS-bashing I've read on blogs and identica/twitter recently. Ironically, some of his most dismissive critics are 20 and 30 somethings who argue that his stance is not convenient; yet it is precisely because of his principles that they enjoy the convenience of a robust free operating system today. And now let me indulge in a grotesque generalization: These are generally the same sorts of people who want to believe that Gnome is every bit as polished as proprietary desktops (which it is not) and that next year, finally, will be the year of the Linux Desktop. They also tend to be more likely to buy into the whole "get with the 21st century", Web Squared, cloud computing, GUI is always better, Lifehacker fan, Google is cool, "give my data to the world" ideology.

RMS is a contrarian. And in this sheepish world we live in, the more contrarians there are, the better.

Re: Package Mangement and Why Your Platform Needs an App Store

tychoish — Wed, 04 Nov 2009 12:37:27 -0000

Shared libraries are what breaks the system. I've been laboring under
the impression, and I'm totally ok being wrong on this, that where we
needed shared libraries to save memory and disk space 10 to 20 years
ago, it's not really an issue now, who cares if there are three
different copies of libxml or libsvg on your system? The benefit you
get from shared libraries is approximately 4 megs, and I have a 4 gig
system...

Certianly this is the approach that Mozilla takes, as they statically
compile GECKO into thunderbird/firefox/sunfire/etc, so that you run four
copies of the rendering engine even for no good reason, except that
they want to avoid breaking things if code assumes a different version
of gecko. Or some such.

And then you have the whole version conflicts if two apps need
different versions of the shared library. And when you have to
managing four different versions of libawesome, because each
libawesome app uses a different version of the library, the benefits
to dynamic linking seem... to be not as beneficial... I'm not
sure. Certainly, static binary packages makes a proprietary package
management system easier from a technological perspective, not that I
think that's a great thing anyway.

I've used Gentoo a bit--mostly testing things for work--and I use Arch
Linux on a day-to-day basis on a number of my machines, and Arch makes
it possible to compile from source as well. Oh, and I used to be a Mac
user and certainly did a lot of port installing then too...

I must say that the "binaries optimized for your CPU" argument doesn't
sit well with me. I mean, I understand that it's true, or can be,
but... I tend to think that the amount of CPU cycles saved rarely
exceeds the amount of CPU cycles it takes to compile things from
source, particularly when--in the case of arch and gentoo--there are
updates released exceedingly frequently. But that's just me. I like
the idea, but I've never gotten to the point where it's seemed
beneficial to me.

Regarding the language-specific package managers, are you familiar
with tools like dh-make-perl? Basically you wget, the tarball of a
cpan module, run dh-make-perl, build the package (debbuild), and then
dpakg -i ___.deb.... It's a pretty sweet system. Not sure how to maek
a similar sort of thing work with ruby gems (which I think is just a
bad system,) nor am I really clear on how to make that kind of awesome
happen with Arch Linux, but I think a little bit of scripts around a
build script...

I dunno. Be in touch.

Cheers,
sam

Re: Package Mangement and Why Your Platform Needs an App Store

tychoish — Wed, 04 Nov 2009 12:00:53 -0000

I've not read a review of a modern UNIX operating system (like a linux
distro) that has made the "finding software" is difficult
argument, but I can definitely imagine how such an article would come
about. The one thing that I imagine that I'm more prone to hearing is
that finding equivalents to proprietary software for Linux is
difficult, which is only sort of true. At least in so far as there are
some very specific applications that people are used to that just
don't exist for Linux yet.

Interestingly, these are the same kinds of applications that don't
exist for OS X either. I imagine that whatever erosion Apple is able
to do of the Windows market share will be good for Linux in the long
run, because once you teach people that they really don't need that
old VisualBasic.NET app from 2001 anymore once... that case becomes
easier the next time.

I'm not sure that the whole App store model is a bad thing, I mean
frustrating, because we've been doing it for a long time with great
success, but at the same time... I think smaller task-specific desktop
applications are a good thing for computer users in the long run. It
would be better if people were using real operating systems--as it
were--but in general the App store "model" seems to be really good for
developers (by giving them a way to monetize application development,
by giving framework developers a way to fund runtimes in a way that
hasn't been possible in a long time, by making the application
development discovery process more coherent.) Like these are all good
things, they could be a lot more free and a lot more open, but...

I think we should be much more wary of the possibility that the next
generation operating systems (like Chrome) won't allow users to
install software *at all*, rather than the possibility that there will
be vendor lock in around a single store (as is arguably the case with
the iPhone at the moment.) Having said that, for all it's frustration,
I think the possibility of ChromeOS being like that... pretty slim,
given google's general business model. It won't be as awesome as Arch,
but then what its?

Re: Package Mangement and Why Your Platform Needs an App Store

madalu — Wed, 04 Nov 2009 10:34:52 -0000

Well, you know that Linux and the BSDs can never win....

Many reviews of Linux distros like make the bogus claim that installing software is difficult. Such claims seems to be based on the belief that it is easier to find an .exe download on the web than it is to browse Synaptic or the like. So the rumor spreads that package management in Linux is difficult, when, in fact, it's so much more convenient than having to rummage around online for software that may or may not be trustworthy.

But when Apple or Android creates a "store" or "market" that is basically just a commodified form of package management, then suddenly it's the hottest thing in the history of computing.

Sometimes I think the biggest problem confronting Linux is that most people can not conceptualize an entire operating system being free.

The Apple App store is in many ways the antithesis of Linux or BSD package management. The latter simply gather together a huge range of open source software available and ensure that they work together smoothly on a particular platform. It is a huge service that the community performs for its members. The App Store, on the other hand, is a highly centralized, highly monetized one-stop shop, which, in the end, places arbitrary limits on what the user can do with the computing device. What happens if desktop OSs such as the vaporware Chrome OS or the rumored Apple Tablet (iPhone) OS, mimick the App Store model, allowing users to install software only via a single, officially-approved channel?

Personal computing as something akin to cable TV service = the vision of the telecoms and Apple.

Personal computing as something akin to broadcast TV (with a lot of advertisements) = the vision of Google.

In either of the above visions, placing limits how software can be obtained and installed helps to create a large, captive audience.

Re: Package Mangement and Why Your Platform Needs an App Store

alaricsp — Wed, 04 Nov 2009 05:34:20 -0000

Yes.

Even on OS X It's quite common for apps these days to use a lot of open-source libraries, though; while a few (openssl, libc, etc) can be assumed as part of the OS, I think there's a growing number of other tools and app might use - I think it'd be a good idea for Apple to, at some point, provide a standard way for apps to require such resources so that the OS can connect them to shared copies of a compatible version. By all means, let the apps come with a copy in their bundle; the installer can detect when multiple apps can all share a compatible version, and then make sure all of them get hardlinks to the same shared copy, that being the most recent version they can all share. Apple have been doing work in the latest version of the OS to try and reduce hard disk and RAM usage, so it might be worth their while looking into that!

Have you ever played with package managers that build from source? I use NetBSD's pkgsrc a lot, which is nice. It takes a long time to compile stuff up, which is the only downside - the upsides being binaries that are optimised for your CPU, often more recent versions, all the packages being available even if you're running on a strange architecture, and the ability to tune compile-time configuration flags to your tastes, easily recompile the package with a patch applied, etc.

Package management in Solaris is painful. While there's a common standard format for binary packages, and a standard registry of what's installed so they can be uninstalled and all that, the package format doesn't cover dependencies, so there's multiple competing "stacks" of packages (often using different compilers and libcs etc) with their own dependency mechanisms. You often need to install more than one (or more than one version of the same one) on the same system, and then get people to set their PATH to select which tools they want to use. *shudder*.

The issue of language-specific package managers such as CPAN is an interesting one. I'd like to see some standardisation, so that tools like CPAN could be asked to export their list of packages and dependencies in some common format so they can be imported into a platform's package manager - so you'd be able to say 'apt-get install cpan:Acme::Thingy' (or have a package depend on that) and it'd know to ask CPAN to install it for you, thereby delegating bits of the package namespace to a federation of smaller-scoped package managers. Or something.

Re: asimov spam

swingtrading — Fri, 30 Oct 2009 01:03:22 -0000

Insightful read. I have stumbled and twittered this for my friends. Others no doubt will like it like I did.

Re: The End of Reusable Software

Basu — Wed, 28 Oct 2009 14:28:26 -0000

Ahh.... the good old days when men were men and wrote their own device drivers. I would like to add that the most powerful pieces of software aren't individual programs, but rather *programming languages*. Case in point: the Processing[1] vs other image processing apps. Photoshop and the Gimp are nice, but the things you can do with Processing are just awesome. Emacs is so powerful because it is essentially a Lisp Machine. From a more theoretical perspective the problem of using a computer is about describing a problem and then either creating a repeatable solution or framing the problem such that the computer can solve it on it's own (using software of course).

In reference to people doing less with computers -- Last year as a research project we tried to make a tool that would allow art students to use Lindenmayer systems[2] to create art. I think it was more or less a failure because the easiest and most powerful way to do is basically to program a Lindenmayer system simulator and the art students refused to learn that even remotely resembled programming. You don't need to know how the machine works, but you need to know how to communicate with it.

[1]http://processing.org/
[2]http://en.wikipedia.org/wiki/L-system

Re: Wordpress Limitations

rupendra99 — Tue, 27 Oct 2009 08:14:15 -0000

it's really helpful,

thanks

Re: five fiction ideas

punksterz779 — Sat, 24 Oct 2009 20:23:12 -0000

wow kewl but the ones about oparah not interested srry

Re: The Tiling Window Manager Story

Basu — Thu, 22 Oct 2009 16:52:58 -0000

I fully understand your comment about Haskell. It's on my to-do list, but it's just far enough out of my current set of language that I'll need to devote serious time to learning it (time I don't have at the moment unfortunately).

As for Lisp, it is a beautiful language (for some definition of beauty). I've been practicing it on and off for the past few weeks and I think that it's best to keep to a mostly functional style, even if Common Lisp does have some imperative-style elements. I think mixing them just makes things messy. As for Emacs, well, in some ways, Emacs is the last of Lisp machines, until they are reborn.

Re: The Tiling Window Manager Story

madalu — Thu, 22 Oct 2009 14:57:36 -0000

I think lisp is *the* language for people who geeked out on 20th century continental philosophy in college. I look at C and think of some Anglo-American analytic philosopher working on symbolic logic. I look at lisp and think: so much verbosity, so many parentheses, such a cascade of statements and clauses, such much freedom in assigning arbitrary variables...

Seriously, lisp is a glorious language. What other language invented in the 50s is still out in front of everything else out there.

As you know, Lisp was designed as an AI language. And while Emacs isn't exactly "intelligent," it is so much more adaptable to the needs of individual users than vim. Emacs lisp is obviously a somewhat old dialect of lisp. But it certainly seems that emacs lisp programmers have been able to achieve a level of intuitive integration that the rest of the computing world is still chasing after. BBDB, gnus scoring, org-mode---all these things are pure magic.

But if you want a universal lisp WM, I still don't understand what's wrong with Emacs. ;)

Re: Operating Systems and the Driver Issue

tychoish — Tue, 20 Oct 2009 10:51:03 -0000

I think the other side of this is that driver support (for anything)
isn't difficult to obtain, you just have to have some measure of
control over the hardware to some degree (this is the apple solution.)
Microsoft (their CA program) and Linux vendors like System76 and
Emperor Linux are good examples of this as well.

My suggestion is that "making better drivers," isn't really a
solution. Standardizing hardware interfaces seems to be a good first
step (and this is part of the lessons, I think, that we've learned
from the whole "cloud thing") but I'm also interested in what's going
to happen next

Re: Operating Systems and the Driver Issue

argosreality — Mon, 19 Oct 2009 21:07:41 -0000

You and I are of course in agreement here. Its a chicken and the egg problem in regards to drive support for Linux. To get good support you need enough user base to get the user base you have to have the driver support!

Also, Microsoft has done alot to change the overall stability of drivers by reguiring drivers to go through proper testing and Q&A before being able to use (legally) the "works with...blah blah" logos and to get marketing dollars. Of course, this doesnt stop shady fly by night POS hardware from stealing the logos and using them but it helps. Maybe if Linux had something similar (and I dont consider a community run website of "well...it sorta works...on my system...only" to be enough) with a large basis of standardized hardware it could help

Re: Simple Beret Pattern (a Free Knitting Pattern)

janefritz — Sun, 18 Oct 2009 12:05:38 -0000

I'm kind of confused with the
In the next round, K3 K2tog around. Knit 5 rounds even in st st
What does knit 5 rounds even in st?
Is it the first round k3 k2tog all the way around and then just knit for 5 rounds in a straight stitch?
And so on...

Re: The Odd Cyborg Out

Wootasaurus — Sat, 17 Oct 2009 00:37:18 -0000

I think I'm the office mate, or at least one of two.

"Oh dear, what's next?" sounds right.

Re: Breaking up with the Web

madalu — Fri, 16 Oct 2009 10:42:59 -0000

Yes, I'm with you. I get downright scared whenever people say that they *want* to do everything in a browser. Web browsers are among the least efficient pieces of software I've used this side of Microsoft Office. That's not the fault of the browsers. Rather it's a product of the logic of the web, which leaves all the user interface in the hands of the web designer/coder. So to access the data on the web, you need to learn a new interface/logic for each website you visit. Plugins like vimperator or browsers like conkeror or uzbl allow the user to seize back some measure of control. But if a website doesn't provide a robust api for accessing it's data through other means, I'm pretty much prepared to write it off.

BTW, that's why I like identica. With xmpp and a rich api, there are lot's of ways to get at the site's data.

I also use emacs-w3m. My favorite part of emacs-w3m by far is shimbun, which allows me to scrape off all the cruft from my favorite websites and read just the content in an emacs mail client (in my case Gnus).

That said, I don't think we're going change things anytime soon. The average computer user is a casual browser who just wants to be entertained and doesn't care about standards or openness. Recently, I've heard several people who should know better suggest that we should abandon email communication for Facebook! (To me, this is like saying we should only ever eat at McDonald's.) My guess is that less than 10% of computer users even know what RSS is (and that's probably a generous estimate).

Re: The Odd Cyborg Out

tychoish — Thu, 15 Oct 2009 14:36:44 -0000

I suspected that you'd say something like that.

Re: The Odd Cyborg Out

madalu — Thu, 15 Oct 2009 14:36:32 -0000

Hmm. Just a few windows. Multiplexed applications. That sounds like a job for Openbox. ;)

Re: The Odd Cyborg Out

madalu — Thu, 15 Oct 2009 14:34:48 -0000

Oh dear...