Anyone know when similar improvements to GTK are coming out?
AFAIK, The font handling in gtk+ 1.2 makes implementing the new scheme painful (although there are some hacks around if you really want it...). Gtk+ 2.0 will have AA support supposedly, with much better font handling all round.
What I'd really like to see is some sub-pixel rendering. Then we could start to see what LCD displays (and other displays with very precise pixel placement) can do.
Already there. Just put "Xft.rgba: rgb" in your x resources and bingo, sub-pixel antialiasing.
Phil
--
Lets test that hypothosis... put a bunch of ice in a cup. Fill cup with water. Hell, just for fun, build a 'city' on the rim of
that cup. Wait for ice to melt.
I'm betting your city isn't flooded, and the cup isn't overflowing...
Note that this argument in turn is flawed; The ice which causes the sea level to rise is that which is currently sat on land masses (eg Antarctica).
I am a linux fan, but have you ever read dot-truth.com? There are links to magazines that tell about the problems of Sun computers. It took forever for them to get ECC ready, while it's been on Intel's for years.
Note that dot-truth.com is really a microsoft site. I doubt its going to be particularly objective about problems with Sun machines!
Getting searching right isn't just something minor, y'know. It's incredibly important.
Indeed. My gf is a librarian + spends a considerable chunk of her time training people to use search engines effectively. (She works in an academic library)
Often the information is out there, but getting to it can require a lot of skill and background knowledge that most people don't necessarily have.
Phil --
You have to add the truetype module to
the list of Xserver extensions in the
XF86Config file, and make sure your truetype
fonts are in the fontpaths declared in that
file too. Finally, the fonts.dir file needs
to exist in the fonts directory (though if you
had it working under 3.3.6 with xfstt then that side of things should be ok....)
Possibly. Bear in mind that the example in question is a monospaced font for a start. And unfortunately truetype fonts vary wildly in the quality of their hinting. Unlike ps fonts (which have a stroke based hinting scheme), tt fonts have a small program included with them which is executed on a virtual machine in order to work out the hinting. Therefore the quality of the hinting for a postscript font is dependant mostly on the quality of the ps renderer, whilst the quality for a tt font can vary wildly between fonts, as it depends on how good the code is for each font.
I believe that part of the reason for MicroSoft's 'web fonts' on screen quality is down to the amount of work they put into the hinting code.
Font nuts usually use this as an argument as to why ps fonts are better than tt fonts.
Phil
ps. Note that none of the above affects the printed quality of fonts, only the rendering on the relative low-res computer displays we currently suffer before... --
Antialiasing at small point sizes is unhelpful on most display technologies as all it does it makes the text blurry - try it, it looks crap. That's why they don't do it.
I'm sorry, but as far as I'm concerned you're so completely wrong its not even funny.Maybe its just a personal preference thing, but I'll happily sacrifice a little 'blurring' if it means I get more readable text plus the ability to actually read text rendered at very small point sizes.
Like Toby (above) {Hi Toby!} I've seen and used the Acorn font renderer and the difference it made to readability and perhaps more importantly useability was outstanding.
Keith Packers example page at
http://www.xfree86.org/~keithp/render/compare.html
has some good textual examples. Note that yes, clearly a hand drawn bitmap font at a particular font size will be fine, but having true antialiased vector fonts gives you so much more flexibility and readability that you don't want to go back.
Even the K6 (Maybe, definately the K6-2 and K6-3) supported SMP.
I don't think this is true. IIRC the story was that the K5 (remember that!) was smp capable but because no-one ever made smp motherboards for it, AMD gave up on SMP for the K6. The Athlon is the first AMD SMP capable processor since the K5.
I may be wrong (so all those astophys grads can correct me if they want) but I don't think Hawking radiation works quite like that.
No, you're right. He's talking about X-ray emmission from infalling matter, which was predicted long before Hawking radiation was. I don't believe anyone has actually seen Hawking radiation yet --- in a black hole of any size it's very small indeed; you only get significant amounts from very small black holes. Indeed, as the black hole looses mass to Hawking radiation, the radiation rate increases (exponentally? Can't remember the power...), this means you should get an explosion of radiation as the black hole evaporates entirely, leaving a naked singularity behind it...
Possibly fortunately for us, the evaporation time for a black hole resulting from a stellar collapse is on the order of 100-1000s of billions of years, and it'll only start after the influx of mass into the black hole falls below the rate of Hawking radiation. This won't happen til the famous heat death of the universe:)
Its possible that micro black holes could be left over from the formation of the universe, and these might evaporate during the lifetime of this galaxy. But no-ones seen it happen yet...(AFAIK, gamma ray bursters have the wrong time-signature to be evaporating black holes)
Obviously, I have something wrong. So how do mirrors that reflect X-rays work?
You're absolutely right that an X-ray photon will rip electrons (even core electons) out of an atom, and will not therefore be reflected. However, at very low angles of incidence (a few degrees at most) X-rays are reflected. (I think this is because they interact with many atoms at once, rather than just one -- quantum effects come in here, and I can't remember the details:( ). Anyway, whats this means is that you can build an 'X-ray lens' by nesting lots of carfully arranged cylinders, whose axis points in the direction you're looking. X-rays reflect off the inner surface of the cylinder onto your detector at the back of the telescope.
This isn't the only way to do it; a slightly weirder method is to put a Uniformly Redundant Array where you would normally put the lens. By selectively subtracting X-rays from the source image, this result in the detector seeing convolution of the image (like a fourier transform) which can be deconvolved to the original image. see here for an introduction to URA high energy telescopes...
URA telescopes like BLAST have the disadvantage that you reduce their sensitivity due to blocking half the incident X-rays. On the other hand, I think you can use them do detect higher energy X-rays than low-angle reflection telescopes. But someone who actually works on these things will probably contradict me:)
Oh come on, anyone who had read HHGTTG at all would have tried to put the fish in their ear. The babelfish puzzle isn't infamous for that end problam at all. For me it's the classic sequence of 'this time I'll be able to stop the babelfish disappearing' moments that makes it a classic.
Just when you thought you'd got it: Oh no, you've just got to the next bit:)
It's certainly true that graphical adventure games like the Sierra (you know who you are Roberta Williams) ones have a lot to answer for, with their total lack of internal logic. Even worse were the 'which pixel do I have to click on to see if this is part of the background or a vital object without which I will be unable to complete the game' games. Aargh!
On the other hand, the Monkey Island games were classics. I still fondly remember the spitting contest...
There's still a thriving community turning out quality text adventures. They're just not commercially viable any more, which is why you never see them mentioned in the 'mainstream' computer games press, who only care about advertising revenue from computer games manufacturers and hence will never look at 'free' games.
What's this about the Gnome project strong-arming the developers into switching from C++ to C? Leaving aside my religious beliefs in OOP, that doesn't seem in keeping with "open
source ideals" at all.
Well, given gnome's cross-platform ideals, C++ can be a bad choice for core libraries. I've seen complaints than Sun changed the ABI for C++ libraries with each of the major releases of their C++ compiler. This is not good if you expect your code + system libraries to interoperate in the slightly longer term. C based libraries have a very stable interface on the other hand which is well established + unlikely to change.
This argument wouldn't matter if we weren't talking about a core API -- you can write an app in whatever language you've got bindings for.
Has anyone written core 'system' type libraries in C++ and made them accessible to other languages than C++? I haven't seen any, but that doesn't mean it hasn't been done...
That would be Lord William Rees-Mogg yes? Oh, look a quick check of the link at the end of the article reports:
The Sovereign Individual: Mastering the Transition to the Information Age
By Davidson, James Dale / Rees-Mogg, William
The problem is, after you get your result, how do you maintain reversibility while reading it
You don't. You loose some energy in outputing the result, the rest can be reversed. Of course, this is in the thermodynamic limit, ie an infinitesinally slow calculation. If you want an answer within your lifetime, you have to spend a bit more energy:)
See something like the "Feynman Lectures on Computation" for a readable summary of this stuff.
Compare this with today's standard laptop, which has a clock speed of about 500 megahertz
and carries out up to 1000 parallel operations each cycle
Where can I get one of these? I thought I was lucky if all three pipelines issued at once in my
processor...
I imagine the author means 1000 parallel single bit operations, given that the rest of the time he is counting single bits. Doesn't sound too outrageous. Think four-way superscalar on a 64 bit architecture + floating point calcs in one clock, and you can easily get over 500 bits of data changing in a single cycle. And we're only really talking order of magnitude stuff here anyway, so its close enough...
Is there no chance of convincing management to go to NDS rather than AD? Novell seem to be *much* happier to support all OS configurations + combinations out there (can't think why:) )
Eh? Whether you can cut and paste stuff is independent of how it's rendered on the screen;
cut and paste is largely done by the client(s), with the only server involvement typically
being the use of the selections such as the primary and clipboard selection.
Exactly. Everyone knows how to do clipboard selection of text in X. At the moment if you move to rendering anti-aliased text you have to add the clipboard support in as well by hand. Of course, a reasonable 'AA text library' would do this for you I suppose....but they'll only use it if and when either the application, or the toolkit it uses, is changed to use the rendering extension, unless they somehow bury that stuff in Xlib itself, which I have the impression they're not doing (and I have the impression might not be doable).
Indeed. I meant that every app will have access to these extensions, without relying on a myriad of client side libraries, all of which do things slightly diferently.
Slashdot Mirror
Hello. How do I turn sub-pixel rendering on?
In your Xresources file (~/.Xdefaults perhaps) add the lines:
Xft.core: 1Xft.render: 1
Xft.scale: 1
Xft.rgba: rgb
You can have either rgb or rbg I think; depending on the layout of your lcd pixels.
Phil (Hoping he's got this right...)
--
Anyone know when similar improvements to GTK are coming out?
AFAIK, The font handling in gtk+ 1.2 makes implementing the new scheme painful (although there are some hacks around if you really want it...). Gtk+ 2.0 will have AA support supposedly, with much better font handling all round.
Phil--
What I'd really like to see is some sub-pixel rendering. Then we could start to see what LCD displays (and other displays with very precise pixel placement) can do. Already there. Just put "Xft.rgba: rgb" in your x resources and bingo, sub-pixel antialiasing. Phil
--
Lets test that hypothosis... put a bunch of ice in a cup. Fill cup with water. Hell, just for fun, build a 'city' on the rim of that cup. Wait for ice to melt.
I'm betting your city isn't flooded, and the cup isn't overflowing...
Note that this argument in turn is flawed; The ice which causes the sea level to rise is that which is currently sat on land masses (eg Antarctica).
Phil--
I am a linux fan, but have you ever read dot-truth.com? There are links to magazines that tell about the problems of Sun computers. It took forever for them to get ECC ready, while it's been on Intel's for years.
Note that dot-truth.com is really a microsoft site. I doubt its going to be particularly objective about problems with Sun machines!
--
Getting searching right isn't just something minor, y'know. It's incredibly important.
Indeed. My gf is a librarian + spends a considerable chunk of her time training people to use search engines effectively. (She works in an academic library) Often the information is out there, but getting to it can require a lot of skill and background knowledge that most people don't necessarily have. Phil--
Has anybody here done that?
Works For Me(tm)
You have to add the truetype module to the list of Xserver extensions in the XF86Config file, and make sure your truetype fonts are in the fontpaths declared in that file too. Finally, the fonts.dir file needs to exist in the fonts directory (though if you had it working under 3.3.6 with xfstt then that side of things should be ok....)
--
Possibly. Bear in mind that the example in question is a monospaced font for a start. And unfortunately truetype fonts vary wildly in the quality of their hinting. Unlike ps fonts (which have a stroke based hinting scheme), tt fonts have a small program included with them which is executed on a virtual machine in order to work out the hinting. Therefore the quality of the hinting for a postscript font is dependant mostly on the quality of the ps renderer, whilst the quality for a tt font can vary wildly between fonts, as it depends on how good the code is for each font.
I believe that part of the reason for MicroSoft's 'web fonts' on screen quality is down to the amount of work they put into the hinting code.
Font nuts usually use this as an argument as to why ps fonts are better than tt fonts.
Phil
ps. Note that none of the above affects the printed quality of fonts, only the rendering on the relative low-res computer displays we currently suffer before...
--
You're right -- My A310 had a 25Mhz ARM3 in it. It got donated to a school eventually. sob
I know, replying to my own post, but before the pedants get to me, I should point out that this was a heavily upgraded A310 :)
Phil--
by the way I seem to remember some of the older ARM3s were at 25Mhz, could be wrong though... I had one put in my A3000 back in the day.
You're right -- My A310 had a 25Mhz ARM3 in it. It got donated to a school eventually. sob
Phil--
Antialiasing at small point sizes is unhelpful on most display technologies as all it does it makes the text blurry - try it, it looks crap. That's why they don't do it.
I'm sorry, but as far as I'm concerned you're so completely wrong its not even funny.Maybe its just a personal preference thing, but I'll happily sacrifice a little 'blurring' if it means I get more readable text plus the ability to actually read text rendered at very small point sizes.
Like Toby (above) {Hi Toby!} I've seen and used the Acorn font renderer and the difference it made to readability and perhaps more importantly useability was outstanding.
Keith Packers example page at http://www.xfree86.org/~keithp/render/compare.html
has some good textual examples. Note that yes, clearly a hand drawn bitmap font at a particular font size will be fine, but having true antialiased vector fonts gives you so much more flexibility and readability that you don't want to go back.
just my 2p :)
Phil--
Even the K6 (Maybe, definately the K6-2 and K6-3) supported SMP.
I don't think this is true. IIRC the story was that the K5 (remember that!) was smp capable but because no-one ever made smp motherboards for it, AMD gave up on SMP for the K6. The Athlon is the first AMD SMP capable processor since the K5.
--
Have a look at http://c2.com/cgi/wiki?WelcomeVisitors for enlightenment :)
--
I may be wrong (so all those astophys grads can correct me if they want) but I don't think Hawking radiation works quite like that.
No, you're right. He's talking about X-ray emmission from infalling matter, which was predicted long before Hawking radiation was. I don't believe anyone has actually seen Hawking radiation yet --- in a black hole of any size it's very small indeed; you only get significant amounts from very small black holes. Indeed, as the black hole looses mass to Hawking radiation, the radiation rate increases (exponentally? Can't remember the power...), this means you should get an explosion of radiation as the black hole evaporates entirely, leaving a naked singularity behind it...
Possibly fortunately for us, the evaporation time for a black hole resulting from a stellar collapse is on the order of 100-1000s of billions of years, and it'll only start after the influx of mass into the black hole falls below the rate of Hawking radiation. This won't happen til the famous heat death of the universe :)
Its possible that micro black holes could be left over from the formation of the universe, and these might evaporate during the lifetime of this galaxy. But no-ones seen it happen yet...(AFAIK, gamma ray bursters have the wrong time-signature to be evaporating black holes)
PhilObviously, I have something wrong. So how do mirrors that reflect X-rays work?
You're absolutely right that an X-ray photon will rip electrons (even core electons) out of an atom, and will not therefore be reflected. However, at very low angles of incidence (a few degrees at most) X-rays are reflected. (I think this is because they interact with many atoms at once, rather than just one -- quantum effects come in here, and I can't remember the details :( ). Anyway, whats this means is that you can build an 'X-ray lens' by nesting lots of carfully arranged cylinders, whose axis points in the direction you're looking. X-rays reflect off the inner surface of the cylinder onto your detector at the back of the telescope.
This isn't the only way to do it; a slightly weirder method is to put a Uniformly Redundant Array where you would normally put the lens. By selectively subtracting X-rays from the source image, this result in the detector seeing convolution of the image (like a fourier transform) which can be deconvolved to the original image. see here for an introduction to URA high energy telescopes...
URA telescopes like BLAST have the disadvantage that you reduce their sensitivity due to blocking half the incident X-rays. On the other hand, I think you can use them do detect higher energy X-rays than low-angle reflection telescopes. But someone who actually works on these things will probably contradict me :)
You're right, I'm wrong. Mea Culpa :)
Oh come on, anyone who had read HHGTTG at all would have tried to put the fish in their ear. The babelfish puzzle isn't infamous for that end problam at all. For me it's the classic sequence of 'this time I'll be able to stop the babelfish disappearing' moments that makes it a classic.
Just when you thought you'd got it: Oh no, you've just got to the next bit :)
It's certainly true that graphical adventure games like the Sierra (you know who you are Roberta Williams) ones have a lot to answer for, with their total lack of internal logic. Even worse were the 'which pixel do I have to click on to see if this is part of the background or a vital object without which I will be unable to complete the game' games. Aargh!On the other hand, the Monkey Island games were classics. I still fondly remember the spitting contest...
There's still a thriving community turning out quality text adventures. They're just not commercially viable any more, which is why you never see them mentioned in the 'mainstream' computer games press, who only care about advertising revenue from computer games manufacturers and hence will never look at 'free' games.
For those who want to relive the infocom experience with new (and some might say, even better!) games, check out ftp://ftp.gmd.de/if-archive, the newsgroups rec.games.int-fiction, rec.arts.int-fiction and wesites like XYZZYnews
Enjoy!
What's this about the Gnome project strong-arming the developers into switching from C++ to C? Leaving aside my religious beliefs in OOP, that doesn't seem in keeping with "open source ideals" at all.
Well, given gnome's cross-platform ideals, C++ can be a bad choice for core libraries. I've seen complaints than Sun changed the ABI for C++ libraries with each of the major releases of their C++ compiler. This is not good if you expect your code + system libraries to interoperate in the slightly longer term. C based libraries have a very stable interface on the other hand which is well established + unlikely to change.
This argument wouldn't matter if we weren't talking about a core API -- you can write an app in whatever language you've got bindings for.
Has anyone written core 'system' type libraries in C++ and made them accessible to other languages than C++? I haven't seen any, but that doesn't mean it hasn't been done...
By Davidson, James Dale / Rees-Mogg, William
*not* impressive.
You don't. You loose some energy in outputing the result, the rest can be reversed. Of course, this is in the thermodynamic limit, ie an infinitesinally slow calculation. If you want an answer within your lifetime, you have to spend a bit more energy :)
See something like the "Feynman Lectures on Computation" for a readable summary of this stuff.
I imagine the author means 1000 parallel single bit operations, given that the rest of the time he is counting single bits. Doesn't sound too outrageous. Think four-way superscalar on a 64 bit architecture + floating point calcs in one clock, and you can easily get over 500 bits of data changing in a single cycle. And we're only really talking order of magnitude stuff here anyway, so its close enough...
Is there no chance of convincing management to go to NDS rather than AD? Novell seem to be *much* happier to support all OS configurations + combinations out there (can't think why :) )
Summit?! Ah ha ha ha! :)
Try Large File support. Oh contraire. You're wrong I'm afraid. The patches are named after the Large File Summit. See http://ftp.sas.com/standards/large.file/ and there's no need to be rude
Eh? Whether you can cut and paste stuff is independent of how it's rendered on the screen; cut and paste is largely done by the client(s), with the only server involvement typically being the use of the selections such as the primary and clipboard selection. Exactly. Everyone knows how to do clipboard selection of text in X. At the moment if you move to rendering anti-aliased text you have to add the clipboard support in as well by hand. Of course, a reasonable 'AA text library' would do this for you I suppose. ...but they'll only use it if and when either the application, or the toolkit it uses, is changed to use the rendering extension, unless they somehow bury that stuff in Xlib itself, which I have the impression they're not doing (and I have the impression might not be doable).
Indeed. I meant that every app will have access to these extensions, without relying on a myriad of client side libraries, all of which do things slightly diferently.