They don't deal with bad weather well. A gusty day will ground most blimps, if they don't get blown away first. Really big airships offer too much surface area to incoming wind in most directions.
Somewhere out there, some old timers once dismissed as "dinosaurs" must be smiling. A flock of 70 boxes got replaced by a single S/390 mainframe. All hail big iron!
Have you noticed that hardly anyone bothers to post memory consumption numbers? Galeon is supposed to be a "light weight" alternative to Mozilla. Apart from the fact that any browser is "light weight" compared to Mozilla, Galeon is still not that small. I've seen "top" figures indicating something to the order of 30MB for Galeon: about the same as Mozilla at that time. I have no idea how to quantify Konqueror's memory footprint, due to KDE's heavy use of shared memory. As for the other browsers, as far as I know, Opera for Linux is still champ in the memory footprint category.
Take a look at the Konqueror screen shot in the article. That looks like Microsoft's Verdana font from their screen-optimized TrueType collection. I compared it to my own browser using Verdana to verify. Keith, the guy doing the anti-aliasing, probably knows full well that you cannot get better screen fonts than to grab Microsoft's fonts. I admit that anti-aliasing actually makes Verdana look better here: at small font sizes like in the screen shot, plain rendering results in scrawny one-pixel lines. The anti-aliasing makes the strokes look thicker, improving their appearance. TrueType is still the best solution, but I think anti-aliasing does not hurt in this case.
Anti-aliasing bad fonts will not make them good fonts. If you want scalable fonts, XFree86 is still not there, at least out of the box.
There are good Type 1 fonts out there, but the font renderer that comes with XFree86 is junk and produces consistently bad results. Besides, TrueType is "by far the technology that produces the best looking monochrome bitmaps", as the Freetype project acknowledges.
But TrueType fonts take real expertise to create. The best free screen fonts you can get are excusively from Microsoft Typography. They are great (Georgia is my favorite), but cannot be freely redistributed. You have to download them from Microsoft. Designing good, well-hinted, screen-optimized TrueType fonts is beyond the ability of most, if not all, open source hobbyists. This means that for Microsoft-haters in the Linux world, the only way to get good screen fonts is through Microsoft, or to buy them.
Type 1 fonts look like crap on XFree86, regardless of the quality of the font or its hints. The reason is that the Type 1 renderer is bad. Even when a better engine appears, TrueType has a natural advantage. Even the Freetype project acknowledges that TT is "by far the technology that produces the best looking monochrome bitmaps". In addition, there are excellent screen-optimized TT fonts available from Microsoft, Monotype and others. There is absolutely no way to get better scalable screen fonts than to use TrueType with these fonts.
I did a "man dhcpcd", copied the text into my text editor, and went to that link you cited to compare the font rendering side by side. Guess what? My non-anti-aliased text -- the same text at the same size -- looked sharper, clearer and better formed. That's because it is using "Courier New", a TrueType font that I snagged from Microsoft's Typography page. Good fonts will beat anti-aliased bad fonts any time.
Another thing that Microsoft does right is to design excellent fonts for the screen, as someone else has pointed out. If you go to the Microsoft Typography page, you will find a bunch of free TT fonts. These are absolutely the best scalable screen fonts you can get anywhere. TrueType fonts can be rendered with quality comparable to hand-designed bitmap fonts. In addition, these fonts are designed especially to look good at low resolution. I use Georgia now for web browsing. I would rather have good screen fonts than antialiased bad fonts.
One possible reason for Java's lack of popularity in the Slashdot world is that it meets rather different needs. Slashdotters like to write little programs: many projects are solo or small scale projects. Browse Freshmeat et al and you will find umpteen different text editors, xbiff clones, command line utilities, GUI front ends for command line utilities, Tetris clones, chat clients etc.
The trouble is that Java is exactly the wrong solution for these apps. The current JVMs are fast enough now, but they all have big memory footprints. Swing especially is a pig. Because much of the Java environment is bytecode loaded at runtime, very little of the footprint can be implemented as shared libraries. Do you really want to run a Java xbiff or xclock clone when every little app you run eat 10-30MB of RAM and take an eternity to start? There is no general solution for the footprint problem, neither now nor for the immediate future. (I know about Echidna, but it is neither usable nor progressing).
I like Java. I write Java code -- including Swing -- for a living, and there are many positive things to say about it. I do not need to repeat its many advantages here. As a server, or as the sole running app in a custom environment it is fine. But I would not want to run a bunch of Java mini-apps on my desktop, at least not with its current state of implementation.
We know all about the demo disk: hard real time, GUI, TCP/IP, web browser all on one bootable floppy. Now try doing that with Linux, a much newer OS. It used to be fashionable to deride Microsoft's "bloated" software, but now look at a similarly configured Linux setup. Linux is now so big that it's not funny. I am not trolling so much as asking what it is that keeps Linux from offering the kind of efficiency that QNX offers. Is there a fundamental barrier? Was Tanenbaum right after all in that famous exchange with Linus?
I know full well the full sentence, but quoting that does not help his case. Firstly, to "take the initiative" means to initiate something. Again, you cannot "initiate" something that already started, namely any effort at creating the Internet. Secondly, the Internet already existed years before any Gore work. Thirdly, Congress did not create the Internet, so no "congressional initiative" could be responsible for its creation.
I am puzzled at the acrobatics Gore defenders do, such as talking about creating something that already exists. Get this: you can't create something that already exists. Sure, it was "far from done" in 1986, but you might as well apply that phrase to 1996 or 2006. It is constantly evolving, but it already exists. And it existed as early as 1983, as even Cerf acknowledge. And this Internet did not suddenly vanish and a new one appear to replace it: we are using the same Internet. Sheesh. Is it so hard to accept the fact that the Internet existed years before any Gore's act of "creation"?
Finally, I am well aware of the Snopes article. But the fact is, it does not even come close to defending the avalanche of nonsense coming from Gore. Al Gore pretty much brought all this criticism on himself: if he could only refrain from lying, none of this would be relevant. As it is, an article touching on a few alleged exaggerations of Gore's exaggerations (delicious irony there) will not do much against the numerous lies on file in his own words since 1984.
He said that that he "took the initative". You don't "take" a "Congressional initative". Clearly, that phrase is meant in the nonlegistlative sense. Moreover, you cannot create what already exists. The Internet Timeline shows that his legistlative work was in 1991, but Vint Cerf acknowledged that "the Internet, as we know it today, was not deployed until 1983". And that is discounting all the work beginning in the 60s. Gore himself did not even start talking about the Internet until 1987. In other words, Gore was not even in picture when the Internet was being created, and was active only years later.
With friends like these, who needs enemies? Look, Gore said "I took the initiative in creating the Internet". Those are his actual words. Yet even Cerf and Kahn's defence of Gore acknowledged that "the Internet, as we know it today, was not deployed until 1983". That's discounting all the stuff done in the 60s. Even this late date cannot rescue Gore. The Internet Timeline indicates that Gore's legislative work took place in 1991. Cerf and Kahn try to salvage whatever they can of Gore's reputation by saying that he was one of the earliest cheerleaders, but that started in 1987: years after the Internet's existence.
In short, Gore's problem is an ontological one: you cannot initiate what has already started, and you cannot create what already exists. Yet this is precisely what Gore claimed, a claim that is so obviously undefendable. Unfortunately, he seemingly cannot stop pulling stuff out of the thin air, even when unpressured. Even at the first debate, he was claiming a nonexistent trip FEMA trip or lack of funding for a well-funded school.
The fact is, one will be hard pressed to defend a lifetime of lies, especially when that entails numerous documented fibs in his own words. The evidence is overwhelming, despite the feeble efforts of his defenders. Gore and Clinton are in a league of their own. What makes Gore worse is that while Clinton lies for self-preservation, Gore unnecessarily lies for self-aggrandization.
We have no idea when that release will see the light of day. Should all C++ development freeze until gcc 3 appears, months or years from now? It's a moot point anyway, since Red Hat 7 has shipped with the incompatible gcc-2.96. Stable C++ binary compatibility remains a dream.
Why not? It's convenient to be able to type, say, "opera" and have the browser append the www and com for me. If you check both options, it should look for a local machine first before appending and prepending those bits. Name completion is turned off by default, incidently. I don't see what you are complaining about. You can make the browser do what you want. It's all about choice.
I get annoyed when I see people talk about how large/small a program is without actually quantifying it. Here, I will do my part. I loaded Netscape 4.75 and Opera 4.0b1 and went to Slashdot (this very article). Their memory footprint can be seen below.
PID USER PRI NI SIZE RSS SHARE STAT LIB %CPU %MEM TIME COMMAND 731 root 15 0 14392 13M 1044 S 0 0.9 21.8 31:48 X 14806 chris 7 0 13164 12M 8088 S 0 0.0 20.8 0:07 netscape-com 883 chris 0 0 12172 11M 2232 S 0 0.0 18.0 22:05 xemacs-X11 14785 chris 0 0 7172 7172 4392 S 0 0.0 11.3 0:09 opera 14786 chris 0 0 7172 7172 4392 S 0 0.0 11.3 0:00 opera 14530 chris 0 0 4424 3740 2608 S 0 0.0 5.9 0:05 kfm 14822 chris 0 0 3724 3724 3148 S 0 0.0 5.9 0:00 netscape-com 654 xfs 0 0 3472 3052 988 S 0 0.0 4.8 1:01 xfs 735 chris 2 0 3252 2276 1708 S 0 0.0 3.6 0:43 kwm 771 chris 0 0 2848 1844 1260 S 0 0.0 2.9 0:43 kpanel 14782 chris 17 0 868 868 664 R 0 2.3 1.3 0:05 top
In short, Opera looks a lot smaller than Netscape, even when statically linked to Qt 2.2. Mozilla usually needs 30MB or so. Anyone care to post numbers for Galeon and Konqueror?
From your perspective, this bill may have a negative in raising the number
of H1B workers. But check the details. In addition to the quota, this bill
improves the portability of H1Bs (the transfer becomes effective at the
time the application is filed) and green card applications (your I-140s
travel with you with no time penalty), increases funding for INS
adjudications and seeks to sharply reduce the backlogs. In other words, it
addresses all your demands except for the part about splitting up INS. But
then again, Bush has pledged to do just that, so just make sure the right
guy gets elected.
There is always this misunderstanding that H1B is only for temporary work.
On the contrary, it is explicitly used as part of the immigration process
for skilled workers. It currently takes years to get an employment-based
green card, so they need a H1B to stay and work in the meantime. That's why
the H1B visa explicitly allows dual intent. More relevant to the topic, this
senate bill would also streamline the green card process for H1B holders, in
addition to cutting immigration backlogs in general.
Some arguments here just miss the point. Some things to keep in mind:
H1Bs are not just for temporary workers. It takes years to get a green card, so qualified workers use H1Bs as an interim measure to stay and work in the meantime.
Forget the slavery argument. H1Bs were portable even before this bill. With this bill, the transfer takes effect at the time of filing, making H1Bs effortlessly portable. In addition, pending green card applications are also portable.
Take a good look at the senate bill S.2045 in question before ranting. Even without this bill, H1Bs can be transferred across companies. With this bill, the transfer takes effect at the time of filing, making H1Bs effortlessly portable. The old "those poor slaves" argument is not relevant here.
Slashdot Mirror
They don't deal with bad weather well. A gusty day will ground most blimps, if they don't get blown away first. Really big airships offer too much surface area to incoming wind in most directions.
Chris
The top output I saw was for a RAM-rich environment. Perhaps you were looking at the RSS figures.
Chris
Each Galeon or Mozilla thread weighs in at about 30MB of RAM. I know better than to add those up.
Somewhere out there, some old timers once dismissed as "dinosaurs" must be smiling. A flock of 70 boxes got replaced by a single S/390 mainframe. All hail big iron!
Have you noticed that hardly anyone bothers to post memory consumption numbers? Galeon is supposed to be a "light weight" alternative to Mozilla. Apart from the fact that any browser is "light weight" compared to Mozilla, Galeon is still not that small. I've seen "top" figures indicating something to the order of 30MB for Galeon: about the same as Mozilla at that time. I have no idea how to quantify Konqueror's memory footprint, due to KDE's heavy use of shared memory. As for the other browsers, as far as I know, Opera for Linux is still champ in the memory footprint category.
Take a look at the Konqueror screen shot in the article. That looks like Microsoft's Verdana font from their screen-optimized TrueType collection. I compared it to my own browser using Verdana to verify. Keith, the guy doing the anti-aliasing, probably knows full well that you cannot get better screen fonts than to grab Microsoft's fonts. I admit that anti-aliasing actually makes Verdana look better here: at small font sizes like in the screen shot, plain rendering results in scrawny one-pixel lines. The anti-aliasing makes the strokes look thicker, improving their appearance. TrueType is still the best solution, but I think anti-aliasing does not hurt in this case.
There are good Type 1 fonts out there, but the font renderer that comes with XFree86 is junk and produces consistently bad results. Besides, TrueType is "by far the technology that produces the best looking monochrome bitmaps", as the Freetype project acknowledges.
But TrueType fonts take real expertise to create. The best free screen fonts you can get are excusively from Microsoft Typography. They are great (Georgia is my favorite), but cannot be freely redistributed. You have to download them from Microsoft. Designing good, well-hinted, screen-optimized TrueType fonts is beyond the ability of most, if not all, open source hobbyists. This means that for Microsoft-haters in the Linux world, the only way to get good screen fonts is through Microsoft, or to buy them.
Type 1 fonts look like crap on XFree86, regardless of the quality of the font or its hints. The reason is that the Type 1 renderer is bad. Even when a better engine appears, TrueType has a natural advantage. Even the Freetype project acknowledges that TT is "by far the technology that produces the best looking monochrome bitmaps". In addition, there are excellent screen-optimized TT fonts available from Microsoft, Monotype and others. There is absolutely no way to get better scalable screen fonts than to use TrueType with these fonts.
I did a "man dhcpcd", copied the text into my text editor, and went to that link you cited to compare the font rendering side by side. Guess what? My non-anti-aliased text -- the same text at the same size -- looked sharper, clearer and better formed. That's because it is using "Courier New", a TrueType font that I snagged from Microsoft's Typography page. Good fonts will beat anti-aliased bad fonts any time.
Another thing that Microsoft does right is to design excellent fonts for the screen, as someone else has pointed out. If you go to the Microsoft Typography page, you will find a bunch of free TT fonts. These are absolutely the best scalable screen fonts you can get anywhere. TrueType fonts can be rendered with quality comparable to hand-designed bitmap fonts. In addition, these fonts are designed especially to look good at low resolution. I use Georgia now for web browsing. I would rather have good screen fonts than antialiased bad fonts.
The trouble is that Java is exactly the wrong solution for these apps. The current JVMs are fast enough now, but they all have big memory footprints. Swing especially is a pig. Because much of the Java environment is bytecode loaded at runtime, very little of the footprint can be implemented as shared libraries. Do you really want to run a Java xbiff or xclock clone when every little app you run eat 10-30MB of RAM and take an eternity to start? There is no general solution for the footprint problem, neither now nor for the immediate future. (I know about Echidna, but it is neither usable nor progressing).
I like Java. I write Java code -- including Swing -- for a living, and there are many positive things to say about it. I do not need to repeat its many advantages here. As a server, or as the sole running app in a custom environment it is fine. But I would not want to run a bunch of Java mini-apps on my desktop, at least not with its current state of implementation.
According to the FAQ, Berlin is 12MB in size. That's still an order of magnitude larger than all of QNX, GUI and web browser on a floppy.
We know all about the demo disk: hard real time, GUI, TCP/IP, web browser all on one bootable floppy. Now try doing that with Linux, a much newer OS. It used to be fashionable to deride Microsoft's "bloated" software, but now look at a similarly configured Linux setup. Linux is now so big that it's not funny. I am not trolling so much as asking what it is that keeps Linux from offering the kind of efficiency that QNX offers. Is there a fundamental barrier? Was Tanenbaum right after all in that famous exchange with Linus?
I am puzzled at the acrobatics Gore defenders do, such as talking about creating something that already exists. Get this: you can't create something that already exists. Sure, it was "far from done" in 1986, but you might as well apply that phrase to 1996 or 2006. It is constantly evolving, but it already exists. And it existed as early as 1983, as even Cerf acknowledge. And this Internet did not suddenly vanish and a new one appear to replace it: we are using the same Internet. Sheesh. Is it so hard to accept the fact that the Internet existed years before any Gore's act of "creation"?
Finally, I am well aware of the Snopes article. But the fact is, it does not even come close to defending the avalanche of nonsense coming from Gore. Al Gore pretty much brought all this criticism on himself: if he could only refrain from lying, none of this would be relevant. As it is, an article touching on a few alleged exaggerations of Gore's exaggerations (delicious irony there) will not do much against the numerous lies on file in his own words since 1984.
He said that that he "took the initative". You don't "take" a "Congressional initative". Clearly, that phrase is meant in the nonlegistlative sense. Moreover, you cannot create what already exists. The Internet Timeline shows that his legistlative work was in 1991, but Vint Cerf acknowledged that "the Internet, as we know it today, was not deployed until 1983". And that is discounting all the work beginning in the 60s. Gore himself did not even start talking about the Internet until 1987. In other words, Gore was not even in picture when the Internet was being created, and was active only years later.
In short, Gore's problem is an ontological one: you cannot initiate what has already started, and you cannot create what already exists. Yet this is precisely what Gore claimed, a claim that is so obviously undefendable. Unfortunately, he seemingly cannot stop pulling stuff out of the thin air, even when unpressured. Even at the first debate, he was claiming a nonexistent trip FEMA trip or lack of funding for a well-funded school.
The fact is, one will be hard pressed to defend a lifetime of lies, especially when that entails numerous documented fibs in his own words. The evidence is overwhelming, despite the feeble efforts of his defenders. Gore and Clinton are in a league of their own. What makes Gore worse is that while Clinton lies for self-preservation, Gore unnecessarily lies for self-aggrandization.
The note says that the gcc steering committee "cannot support" gcc 2.96. So? The gcc team are not in the support business. Red Hat/Cygnus is.
The idea comes from one of the gcc developers. See the posting on LWN.net.
Besides, this is only a beta version of Opera.
Why not? It's convenient to be able to type, say, "opera" and have the browser append the www and com for me. If you check both options, it should look for a local machine first before appending and prepending those bits. Name completion is turned off by default, incidently. I don't see what you are complaining about. You can make the browser do what you want. It's all about choice.
11:32am up 4 days, 59 min, 4 users, load average: 0.65, 0.52, 0.33
67 processes: 66 sleeping, 1 running, 0 zombie, 0 stopped
CPU states: 3.3% user, 1.3% system, 0.0% nice, 95.3% idle
Mem: 63032K av, 60300K used, 2732K free, 29840K shrd, 1308K buff
Swap: 64476K av, 15704K used, 48772K free 21484K cached
PID USER PRI NI SIZE RSS SHARE STAT LIB %CPU %MEM TIME COMMAND
731 root 15 0 14392 13M 1044 S 0 0.9 21.8 31:48 X
14806 chris 7 0 13164 12M 8088 S 0 0.0 20.8 0:07 netscape-com
883 chris 0 0 12172 11M 2232 S 0 0.0 18.0 22:05 xemacs-X11
14785 chris 0 0 7172 7172 4392 S 0 0.0 11.3 0:09 opera
14786 chris 0 0 7172 7172 4392 S 0 0.0 11.3 0:00 opera
14530 chris 0 0 4424 3740 2608 S 0 0.0 5.9 0:05 kfm
14822 chris 0 0 3724 3724 3148 S 0 0.0 5.9 0:00 netscape-com
654 xfs 0 0 3472 3052 988 S 0 0.0 4.8 1:01 xfs
735 chris 2 0 3252 2276 1708 S 0 0.0 3.6 0:43 kwm
771 chris 0 0 2848 1844 1260 S 0 0.0 2.9 0:43 kpanel
14782 chris 17 0 868 868 664 R 0 2.3 1.3 0:05 top
In short, Opera looks a lot smaller than Netscape, even when statically linked to Qt 2.2. Mozilla usually needs 30MB or so. Anyone care to post numbers for Galeon and Konqueror?
From your perspective, this bill may have a negative in raising the number of H1B workers. But check the details. In addition to the quota, this bill improves the portability of H1Bs (the transfer becomes effective at the time the application is filed) and green card applications (your I-140s travel with you with no time penalty), increases funding for INS adjudications and seeks to sharply reduce the backlogs. In other words, it addresses all your demands except for the part about splitting up INS. But then again, Bush has pledged to do just that, so just make sure the right guy gets elected.
There is always this misunderstanding that H1B is only for temporary work.
On the contrary, it is explicitly used as part of the immigration process
for skilled workers. It currently takes years to get an employment-based
green card, so they need a H1B to stay and work in the meantime. That's why
the H1B visa explicitly allows dual intent. More relevant to the topic, this
senate bill would also streamline the green card process for H1B holders, in
addition to cutting immigration backlogs in general.
See the other post and bill summary for more details.
Take a good look at the senate bill S.2045 in question before ranting. Even without this bill, H1Bs can be transferred across companies. With this bill, the transfer takes effect at the time of filing, making H1Bs effortlessly portable. The old "those poor slaves" argument is not relevant here.