I've spent a few hours this week perusing the Imlib source code (I want to add my own image format, like the multigradients in WindowMaker, for use in GTK themes)
Image compression, etc. is only of importance when an image is first being loaded - after that point all the image data is cached, essentially in 24bit color uncompressed.
If you're on a slow system, there might be some lag when pixmaps have to be resized - basically, when GTK needs a pixmap for a button or window of a size that hasn't been requested before, Imlib won't have a cached pixmap of that size and will have to rescale one on the fly. The default rescaling algorithms are pretty nice quality, too; I think you can set Imlib to use cpu-cheaper ones.
If the lag is really Imlib's fault, try running imlib_config, increasing cache size if you have the RAM to spare, and tweaking things like that.
There are more likely possibilities, though:
GTK+ is supposed to be a fairly lightweight toolkit, but I don't know how optimized it is with themes to consider. When I hot-key flicker between desktops running GTK+ apps, for example, all the pixmap backgrounds come up instantly, but there's a perceptible delay for text to be written on top of them.
XFree86 is the second most inefficient piece of software (next to Netscape) on a typical Linux system. This is in no way an attack on the XFree developers; they're coding a system that is at the same level of complexity as the kernel, with a lot less publicity and thanks. Short of avoiding poorly accelerated video cards or waiting for XFree86 4.0 (with DLL-based hardware support, revamped XAA, and lots of other goodies) I don't know what you could do about this one.
I wish the XFree86 people would provide anonymous CVS access. They bemoan a parcity of developers, but they're hardly doing much to attract new blood when all the cool new work is being done behind closed doors. Most latecomer developers to a project don't suddenly wake up one morning and think, "Gee, I want to pledge a massive contribution of time and effort to project X," (pun intended), they instead start by downloading code as bleeding edge as possible to make a few little hacks and tweaks on while getting to know the system. Having to wait months at a time between software releases, with nothing but rumors publicly available inbetween, isn't conducive to this kind of tenative engagement.
They have their reasons, of course; I just think those reasons are outweighed by the benefits of more public development.
Wow, that was a nice rant. Hope I get more "Interesting" than "Offtopic" votes; both probably apply.
The answer to your question is: file format loading times only matter to Imlib once, when the file is loaded. Efficiency while Imlib is actually rendering pixmaps is file format independent.
I think it's/etc/mail/aliases to configure that, but I could be wrong.
You'd best read your root mail somehow - cron misfunctions, or people warning you about problems with your system, are often things you don't want to ignore.
Unix users seem to have a sense of invincibility based on Unix's invulerability to boot sector viruses, floppy viruses, and similar things that require a simple OS kernel and an "every user is root" security model.
That invulnerability doesn't apply to worms (like this, like Melissa). All you need for a worm to work is a homogenous network environment to infect and an exploit to use for the infection. Maybe Unix users are really more savvy and won't fall for trojan horses (the easy "exploit"), but there was a worm created that spread via the imapd hole last year, and any similar exploit allowing so much as a "nobody" shell to be opened on your system could be used for the same purposes.
Do you know what services are running on your Linux box, and have you shut down the ones you don't need? Do you subscribe to bugtraq, redhat-watch-list, or whatever security mailing list is kept up for your distribution?
These were good ideas before, to prevent single crack attempts when exploits were found. Now they're much more important good ideas, as any cracker above the "script kiddie" level is going to be using self-propagating code to start forest fires of attacks.
Maybe the majority of those attacks will be stupid "email attachment" worms like those currently plaguing Windows, and thus incapable of harming system files... but if someone exploits the backticks in/etc/mailcap to delete $HOME, how much better are you going to feel because/usr was untouchable?
For school & work Linux systems I created a preconfigured freshrpms package which includes a cron job to regularly check the redhat errata, download any updated packages, and mail root when something new appears. It's a step in the right direction - Linux is a secure system because bugs are so quickly found and fixed, but it won't be publically perceived as a secure system if security-unconscious newbies never see or apply those fixes.
I suspect redistributing the Buffy episode may be illegal, and I'm certain it's pissing off the WB, but it's hardly immoral.
Don't get me wrong - I believe in intellectual property, I even believe license writers should be allowed to enforce any and every wacky fascist provision they can sucker a customer into agreeing to - but those are cases based on mutual agreement. If you don't like a software license, don't buy the software. If you think Hollywood is ripping you off with $7 movies, don't buy a ticket. If you're pissed that you can't photocopy or web publish a copyrighted book, write your own content. And if you distribute warez, whether it's commercial software or an mpeg of The Matrix, you are basically distributing stolen property.
I'm much less concerned about the morality of republishing a TV broadcast. There's no agreement, stated or implicit, made by someone whose only activity was to intercept radio waves streaming through their own homes and save a copy of the data. There is an exchange which occurs with every other distribution of copyrighted data (hell, even with cable & encrypted satellite broadcasts technically); an exchange that just doesn't occur with broadcast media. If WB wants to retain control over their TV programs, they should have asked us before sending those programs to millions of households.
Let's all start making our emails seven-bit-clean with "red flag" armor - choose, say, 65536 words like "assassination" or "echelon" that are certain to trip email scanning bots, then map every 16 bits of an encrypted message to the red flag word at that index in the word database.
The irony of thousands of innocuous messages, both encrypted and tailored to fill NSA "suspicious message" databases, would be amusing.
If the frame rates on this thing are in the ballpark with the Voodoo 3 3000 (and last I heard they weren't much slower, even with 32 bit rendering on), I'm getting one. I'm worried about performance on my K6II, though - all the benchmarks I've seen show 3dfx cards performing almost evenly with Celerons at the same clock speed, while TNT/TNT2 performance falls off dramatically.
I know a couple other people whose only gripe with the TNT2 was the lack of Linux support - I expect their minds are made up solidly now.
There's a little more info in the Bugtraq post.
on
Linux 2.2 DoS Attack
·
· Score: 5
Piotr Wilkin (pwl@WOTAN.2SLO.WAW.PL) Tue, 1 Jun 1999 17:43:17 +0200
Messages sorted by: [ date ][ thread ][ subject ][ author ] Next message: Salvatore Sanfilippo -antirez-: "whois_raw.cgi problem" Previous message: aleph1@UNDERGROUND.ORG: "New Allaire Security Bulletin (ASB99-09)"
I'm sorry if this has been noticed before, but since I did't find anything in the archives, I post it here. There seems to be a bug in kernels 2.2.x (tested on 2.2.7 and 2.2.9), that causes them to panic when they are sent a large number of specific ICMP packages. I think the problem comes from the combination of the mangled header length (shorter or longer ihl's don't cause hangup) and the random ICMP packets (random type/subtype and source address) this program sends. Windows 9x and FreeBSD 3.0 seem to be unaffected.
I think the most interesting thing is the date, though... I'm sure I'm making a timezone mistake here, but isn't that 8 hours ago? Is that faster or slower than the Linux teardrop fix?
It's annoying to find out about a new DOS attack, but the resolution is all that you could hope for.
It's a little less annoying that there don't seem to be any outstanding instant-crash attacks against Win98 to laugh about - they finally fixed the series of attacks that crashed 95 for 8 months straight, and I haven't seen anything since. Did Microsoft finally get their IP stack right?
If you want to use files larger than 2 GB (the largest number that will fit in a signed 32-bit integer), then get a 64-bit system to put them on. If you want to have simple, efficient, easy to code and easy to port seeks within your files, then you're going to want to be able to use a signed integer to seek back and forth (let's not even mention the trouble with mmap() if your files are larger than a pointer can index...)
The *last* thing Linux should do about 2GB files is try and use hack after kludge to satisfy people who want to use Intel chips but don't want to hear about their limitations.
Because the second time I watched The Matrix, I caught the half-sentence that had escaped my (and apparantly everyone else's) attention earlier: (not an exact quote): "with this bioenergy, plus a cheap form of cold fusion..."
In other words, they're getting gigawatt after gigawatt out of ocean water; the "human battery" thing is just some twisted thing that evil computers do for kicks.
Of course, I would have LOVED it if they had thought of another poster's suggestion: that the apocalyptic world was entirely humanity's fault, and that the Matrix was a "Zeroth Law" reaction of the computers, meant to keep humanity happy and prevent them from hurting each other, for our own good.
It would have lost the "pure evil" factor, but it would have allowed a lot of cool complications to develop in a four-sided war:
The computers who want to help humanity vs. The humans who want freedom at any price vs. The humans who want to cooperate to produce virtual utopia vs. The computers who are sick of supporting inferior life and want to wipe out the whole project to make room for themselves.
I know I post this every few months...
on
PDA+MP3 Player
·
· Score: 4
But I don't want a Palm Pilot, I don't want a Rio, I don't want any of the one-trick gadgets on the market today.
What I want is a portable gadget with wireless 'net access, a built in cell phone (preferably embedded in the PPP stream so I don't have to log out to call someone), sound quality good enough to play MP3s, image quality good enough to run snes9x (even if only in black & white), CPU speed to do both, hardware open enough to run Linux on, a 10baseT port instead of some stupid serial-based "cradle" for uplinks, a set of long-life, built in rechargeable batteries, a fold-out keyboard, a stylus, and a couple PCMCIA slots. (an optional PCMCIA-sized GPS receiver or 4 GB drive would be nice too) Oh, and *lots* of RAM. At a dollar a meg, putting 32 or 64MB RAM into a handheld device isn't ridiculous. I don't know what default software should be included, but at the very least I should be able to look at the bundled math program and throw away my TI-85 (or install Linux/matlab and throw away my TI-85).
This would probably cost over a thousand dollars and require daily recharging with moderate use today, but it is possible, and those numbers are only getting rapidly better. I guarantee the first company to replace the PDA, pager, cell phone, game boy, walkman, calculator, GPS receiver, etc. with one single unit, and make it affordable, will be rich overnight.
What happened to the other problems with FTL travel?
I'm thinking specifically here about the more slippery definition of "simultaneous" in special relativity, and it's consequences. Assume a ship that can:
a. Accelerate to any sublight speed (and thus any "normal" reference frame) arbitrarily fast. b. Use "Warp Drive" to move between two points in space time faster than light.
That's all you need, and assumption a. is perfectly in accordance with physical laws (and technically plausible with some externally powered propulsion system).
Throw in the cute fact that for any two points in spacetime that are not in each other's light cones, there is an inertial frame of reference where those two points are simultaneous, and your ship can:
1. Warp 1000 light years away, simultaneous with the frame of reference of the solar system. 2. Accelerate to near light speed. 3. Warp back to earth, but 999 years before it left.
I wish I could link in a diagram of this... but go look in a physics textbook; it's a classic paradox meant to show why FTL travel is impossible.
I've never heard a good explanation of how Alcubierre's theory (not to mention whatever new concept has come up) deals with this.
I'm surprised nobody seems to have mentioned this so far: With an emulator, I may be able to illegally play pirated games... But I am also able to legally play games which I have purchased, even though I don't own the console system they were designed for.
This is something Nintendo wants to stop, every bit as much as software piracy. It probably hurts their revenues just as much (how many people would pay $200 for a console system that only plays games the $2000 computer they own plays better?), and since it's a completely legitimate practice they can't attack it. I don't own an N64, but if I could get my hands on UltraHLE I'd be happy to borrow a friends Zelda64 cart, download the ROM, and play through it on my computer, deleting the ROM when I'm done. I bet it would look better on a Voodoo 2 than a low-res TV set, too.
The only thing Nintendo can do is change the subject - pretend that because emulators can be used to pirate games, they must only be used to pirate games. Claim that the data on a game cart is somehow an integral part of the cart, not software to be used as the legal owner sees fit. Even claim that their consoles aren't legally computers, and aren't subject to the same rules restricting software licenses.
If I transfer the contents of my hard drive to a faster model, then throw the old drive in the garage, I'm not violating any software licenses. If my old hard drive crashed irreparably and I get copies of my own software from a friend, again I'm doing nothing immoral.
And if I buy a N64 cart, then decide to play it on a quality 3d card/monitor rather than on an N64/grainy TV, even if I never buy an N64 in the first place, that's OK with me. I'm doing nothing wrong, the emulator authors didn't do anything wrong.
The only people in this scenario doing anything close to wrong is the N64 manufacturer attempting to boost sales by artificially maintaining a software lockin.
I thought they might be forming a group to do the only thing that might hurt Linux: make Windows NT efficient, rock stable, compatible with decades of portable code, truly multiuser, secure, and network transparent.
Caldera Open Administration System, that is. The article gave it a passing mention.
It's been in quiet development for at least a year now, Linux Journal claims it's going to be modular, GUI based & vi-compatible, it looks like the only serious competitor to Linuxconf, and it fills the software hole (user friendly & newbie friendly system configuration) that Linux needs most desperately.
Linux is currently at a state where any PC or Mac user could switch to and use it, as long as they had some guru to log in as root whenever mucking around in/etc is called for. I'm hoping we'll see enough improvement with Linuxconf or COAS that Joe Avg. Macuser will be able to handle those tasks as well.
A couple more things I'd like to have cleared up:
Is COAS under the GPL like they said it would be? What non-free software (if any) is on the Caldera 2.2 CDs?
Petreley mentions having to muck about with/etc/ld.so.conf "when adding new libraries" - does he just mean new libc5 libraries, or would I really have to futz around pointlessly every time I make a semi-weekly upgrade to some bleeding edge libs? If it's the latter, count me out. Red Hat's been supporting libc6 & libc5 programs side by side for a year and a half now without that kind of kludge, and all the trouble I've ever had to take was run "ldd" on libc5 programs to make sure all the libraries they want are in/usr/i486-linux-libc5/lib
The best solution, of course, would be if Corel recompiled WP8 (or hurried up on their 2000 product) so I could ditch libc5 compatibility entirely.
I agree. Although I think it is strange that you have not mentioned any lisp dialect here. Guile had been perfect.
I'll agree in theory - although I personally am not a comp sci major and haven't been forced to wrap my brain around functional programming. The idea of say, summing a list via recursion instead of a loop just doesn't feel natural.
I also don't know enough about Guile specifically - how easy is it to interface with code written in other languages?
And BTW, you don't really mean that you prefer writing AI in C?
Just an example, just a theoretical example... but one point I was trying to make is that if you can link to C code, it's possible to write wrappers as a "gateway" to just about any language out there for people who really hate your particular scripting choice.
And not a good one either - what were they thinking? While SLIC does not directly support user defined functions, there is a workaround. I'm sorry; I'll get by without object oriented programming fine, but at least give us functions to work with. I can't even imagine writing a script longer than a couple hundred lines in this. Not to mention the undeclared variables (what do you mean there's a typo in my script? There was no parsing error!), and the lack of locally scoped variables, structures, pointers/references, or many other features of every useful language created in the last 30 years.
They would have had less programmer work, probably earned greater performance, and definitely made the game infinitely more extensible, if they'd added a Perl, Python, or even Tcl interpreter instead of creating this travesty. (Hell, Java, Javascript, etc. would work too if the licensing was OK.)
They also would have tapped in to the vast mountain of existing code out there. Imagine a python script that, at the end of every turn, updates a web page with a summary of the game. Imagine a perl script with the Gimp module, allowing you to make animated GIFs of the battles when you press a button. Imagine writing (and compiling) your own AI in C, then hooking it up to libtcl, perl XS, or whatever...
Not that this is all bad - I mean even crippled user extensible scripts are definitely preferable to no user extensibility at all... but they hit on an idea that could have been as revolutionary to strategy games as QuakeC was to 3D shooters, and it hurts to see they missed the target.
Most of this can be explaned as the use of MP3s for pretty much all of the sounds in the game. They're very processor-intensive little buggers.
That's actually not quite true - anything Pentium or better should have more than enough CPU cycles for decoding MP3s - the problem is that MP3 playing is both somewhat CPU intensive and real time, which means if your MP3 player doesn't have special priviledges and goes 10ms without running, it's going to skip.
Off topic: does anyone know which MP3 players for Linux can do a sched_setscheduler() when they're suid root? x11amp 0.6 did this, and on my system it made the difference between occasional skips and unskippable music - but x11amp 0.6 is getting old.
Anyway, Centauri should still have more than enough CPU cycles left over to animate a small sprite or two quickly, even if every sound effect was being decoded on the fly (and if they're not caching wav data of the small sound effects, they've got a coding problem right there). It honestly feels like the whole game is being run in an interpreter, like it's some Tcl/Tk or uncompiled BASIC app. That doesn't sound likely, but I can't think of anything else that might be sucking up that much CPU.
If they do get a Linux port out, I hope they take the time to fix their atrocious speed problems. Don't get me wrong, it's a great game... except that it was running slower on my friend's PII-400 than it should run on a P166. I'm sorry, there's no reason that popping up an icon menu or scrolling across a 2D map should be anything less than instantaneous on a modern computer.
The lack of a Linux port (when Civ:CTP has one in the works) was one strike against my buying Alpha Centauri, but the fact that Centauri would be unplayable on half my friends' and relatives' computers was just as bad.
Let me get something clear: I don't expect Microsoft to ever do anything more than say "You can order most of our Windows 2000 source code for $200 as long as you sign this agreement not to redistribute it or any patches to it except to us, not to work on any other operating system or Windows API related projects, and to give us your eternal soul." That kind of "new type of open source" would have little or no effect on Windows itself - except it would make life much easier for warez dudez who want to make trojan software and hackers looking for remote exploits.
What I'm worried about is that some clueless but well-meaning judge might decide that the best way to crack down on Microsoft would be to whip out the "eminent domain" stick and take away Windows by making it true open source. That would be OK for free software in general, but for the state of computing it would be a horrible step backward.
I don't just want free software, I want free, good software. We don't need hackers poring over Windows 2000 in some vain attempt to make remote display of apps stable and scalable, or trying to fix whatever hideous kludges they've got preventing them from throwing away DOS finally.
For those people who say that "we haven't seen the source code, how do we know it sucks", we've got enough good evidence to go on. Their repeated attempts and repeated failures to get rid of the 16 bit code and DOS core in Win9x is one example. The travesties of the Win16 API and WINS is another. The fact that they've had a half-baked 32-bit port of NT to the Alpha instead of writing 64-bit clean code in the first place is a third. How about the WinNuke/Teardrop/Syndrop/Newtear/etc series of attacks on the Windows IP code, where the patches took weeks or months to be released, and where half of the new attacks were just minor changes (e.g. using UDP instead of TCP) on the "patched" attacks?
I'm told that NT has some good kernel design decisions, and that it's just the wacky video drivers and the mess of the Win32 API weighing it down... but I'm told this by people who still claim that NT is a "microkernel" architecture.
I'm also personally impressed with OLE, which I think is one example of something Microsoft did themselves, did without too much prior art (All I can think of is ToolTalk in the Unix world - apologies given and corrections requested if I'm missing some Mac history here), and did well... but the underlying implementation isn't of value here (and is probably inferior to CORBA if they're just making it network transparent now), the API (which we already have and which Gnome is practically cloning AFAIK) is.
In short, it's evident that with the exception of accelerated Wine development, an open source Windows would be fairly valueless.
What an open source Windows would do is something I'd rather not see happen: insure the immortality of Windows. More than ensuring "World Domination", open source software is ensuring that free Unix will never die... and for both obvious and personal reasons I hope Windows does die. In the next 5-10 years, please. The idea of Windows dying may have seemed ridiculous a couple years ago, but with Linux pulling out the rug from under NT server and threatening the desktop, with the Godzilla movie of Windows 2000 fast receding, with 64 bit NT development headed by the guy responsibile for Windows 3.1 (which scarred me emotionally), and with even Apple clawing their way back out of the grave...
One of the best things about Unix is that, incompatible vendor extensions aside, all the Unices are built around a set of common, well-designed APIs that give the word "standard" real meaning. I'd hate to see Windows become a standard as well, not because of any design features but because people got to make a lot of cheap copies.
Forget about Linux distro's.... Every damn system is different!
Damn straight. I'm using bleeding-edge releases of my kernel, the C & C++ libraries and compilers, and a few other things at home that I won't be installing at work until they've been hammered on a while. My system here is using Window Maker, at work it's FVWM2, and my roommate is using KDE. He also drives a different car, wears different clothing, and likes different food. It's called "freedom of choice", and we like it.
For the record, all of the systems we own or run execute the exact same software binaries fine, (and the LinuxPPC systems run the same software code fine after a recompile).
We got libc5 libc6.
You forgot libc4. That's right, we've got old systems that only run libc5 binaries, ancient systems that only run libc4/a.out, new systems built around libc6, and new systems that run all three. We've got the same programs recompiling with few or no source code changes on the newest systems, but getting the technical benefits nonetheless. Read up on the technical differences between C libraries, and try to stretch your mind to encompass the concepts of "progress" and "backwards compatibility" at the same time. Once you've grokked that, email Microsoft and ask them why most of their drivers, DOS, and Win16 programs don't work with Windows "New Technology". Then ask them why most of their Win32 programs still don't work with "Windows 2000" without source code changes. Then ask them what's going to happen to that 32-bit specific API when they finally push 64-bit NT (a separate project from Windows 2000...) out the door three or four years from now.
Different widget sets, window managers,
Yeah, yeah: "Ein MFC, Ein Windows, Ein Microsoft." Great world for everyone to be forced to live in. Different widget sets and different window managers fill different needs of programmers and users, accomodate wider sets of preferences, and in general foster competition and evolution of software. Anyone horribly confused by a GTK program sitting on the same desktop as (or an adjacent virtual desktop to) a Lesstif program should probably just give up entirely and get WebTV.
hell even different windowing systems (MetroX, XFree86 et al)
If you think those are different windowing systems, you need to go back and do some more reading. The exact same programs run on either and can't tell the difference. Most hardware works with either happily. In fact, chew on this: not only are different X servers not incompatible, they are so compatible that I can run decades old HP-UX programs on the same desktop next to new Linux software, with the only noticeable change being that Qt and GTK look better than Motif.
A lot of boxes run kernels so well tuned you can't even boot it on another machine!
Yes, boxes whose owners have chosen to make those tuning changes themselves. Nobody's holding a gun to your head to force you to use 486 or Pentium II specific instructions, or to get rid of your EIDE drivers on a SCSI-only machine, you know.
I guess is too bad it all works, after all thats what really counts.
Try not to sound too disappointed.
I really don't understand the backlash here - for most people, if Linux isn't for you, don't worry about, ignore it in the papers and the trade rags, and it won't bother you again until they start advertising idiot proof distributions in 2005. If you're a Win9x user and happy that way then Linux just isn't a factor.
If you're an NT server programmer watching your world start to crumble, on the other hand, I can recommend several good books on the POSIX standard that you may find of interest in the near future.
Currently, the only license information I can find distributed with the kernel itself is the GPL, in/usr/src/linux/COPYING. However, there seems to be the exception, spoken by Linus himself, that binary kernel modules are allowed as long as they don't require interface changes in the kernel proper. We've already had one such case -/dev/3dfx
If this is to be the case, we had better get it in writing and get it in writing ASAP - I can just picture the chaos when binary-only drivers start becoming common, and some kernel developer says "No, I contributed to the linux kernel under the GPL. Stop allowing binary modules, or take my 50K lines of code out of the kernel."
Is that it takes three ideas (a pure OOP language, a portable virtual machine & byte code specification, and a nice GUI library) with lots of promise, then it locks those ideas irrevocably together. If developers could compile to native code, link to existing object files, etc. Java would have a lot broader usefulness. Good job, Cygnus.
The argument essentially is "If a hacker reinstalls your operating system, and they have the source code, they can plant trojan programs." This seems to be the computer equivalent of "If a murderer is hiding in your kitchen, and you own steak knives, they can stab you when you come to breakfast." Technically true, but missing the point entirely. If someone is trojaning your Unix, then they already have root, and you've already been cracked - it doesn't much matter what tools they use to make use of that crack.
He also argues that closed source makes this impossible because they can... er... put up really distinctive splash screens... no, wait, that's not enough... dialog boxes! They can put up distinctive dialog boxes so that the users will see the dialog boxes and know they haven't been hacked.
Would someone please send this guy a trojan Solaris login binary or a copy of Back Orifice to look at? Hell, with the obvious emotional investment and feelings of martyrdom he's showing, even a VB app that puts up the Windows login dialog might be enough to shatter him psychologically.
My personal preference would be to have (Threaded, Sort by Score) pages sort by the highest score in each thread, then display the scores for followups next to the followup names. This would have the disadvantage that the largest threads would invariably filter to the top, but it would mean the people who want to read all the Score:4+ posts and skim everything else (which is why we're sorting by score anyway) would be able to do so most easily.
Slashdot Mirror
I've spent a few hours this week perusing the Imlib source code (I want to add my own image format, like the multigradients in WindowMaker, for use in GTK themes)
Image compression, etc. is only of importance when an image is first being loaded - after that point all the image data is cached, essentially in 24bit color uncompressed.
If you're on a slow system, there might be some lag when pixmaps have to be resized - basically, when GTK needs a pixmap for a button or window of a size that hasn't been requested before, Imlib won't have a cached pixmap of that size and will have to rescale one on the fly. The default rescaling algorithms are pretty nice quality, too; I think you can set Imlib to use cpu-cheaper ones.
If the lag is really Imlib's fault, try running imlib_config, increasing cache size if you have the RAM to spare, and tweaking things like that.
There are more likely possibilities, though:
GTK+ is supposed to be a fairly lightweight toolkit, but I don't know how optimized it is with themes to consider. When I hot-key flicker between desktops running GTK+ apps, for example, all the pixmap backgrounds come up instantly, but there's a perceptible delay for text to be written on top of them.
XFree86 is the second most inefficient piece of software (next to Netscape) on a typical Linux system. This is in no way an attack on the XFree developers; they're coding a system that is at the same level of complexity as the kernel, with a lot less publicity and thanks. Short of avoiding poorly accelerated video cards or waiting for XFree86 4.0 (with DLL-based hardware support, revamped XAA, and lots of other goodies) I don't know what you could do about this one.
I wish the XFree86 people would provide anonymous CVS access. They bemoan a parcity of developers, but they're hardly doing much to attract new blood when all the cool new work is being done behind closed doors. Most latecomer developers to a project don't suddenly wake up one morning and think, "Gee, I want to pledge a massive contribution of time and effort to project X," (pun intended), they instead start by downloading code as bleeding edge as possible to make a few little hacks and tweaks on while getting to know the system. Having to wait months at a time between software releases, with nothing but rumors publicly available inbetween, isn't conducive to this kind of tenative engagement.
They have their reasons, of course; I just think those reasons are outweighed by the benefits of more public development.
Wow, that was a nice rant. Hope I get more "Interesting" than "Offtopic" votes; both probably apply.
The answer to your question is: file format loading times only matter to Imlib once, when the file is loaded. Efficiency while Imlib is actually rendering pixmaps is file format independent.
I think it's /etc/mail/aliases to configure that, but I could be wrong.
You'd best read your root mail somehow - cron misfunctions, or people warning you about problems with your system, are often things you don't want to ignore.
Unix users seem to have a sense of invincibility based on Unix's invulerability to boot sector viruses, floppy viruses, and similar things that require a simple OS kernel and an "every user is root" security model.
/etc/mailcap to delete $HOME, how much better are you going to feel because /usr was untouchable?
That invulnerability doesn't apply to worms (like this, like Melissa). All you need for a worm to work is a homogenous network environment to infect and an exploit to use for the infection. Maybe Unix users are really more savvy and won't fall for trojan horses (the easy "exploit"), but there was a worm created that spread via the imapd hole last year, and any similar exploit allowing so much as a "nobody" shell to be opened on your system could be used for the same purposes.
Do you know what services are running on your Linux box, and have you shut down the ones you don't need? Do you subscribe to bugtraq, redhat-watch-list, or whatever security mailing list is kept up for your distribution?
These were good ideas before, to prevent single crack attempts when exploits were found. Now they're much more important good ideas, as any cracker above the "script kiddie" level is going to be using self-propagating code to start forest fires of attacks.
Maybe the majority of those attacks will be stupid "email attachment" worms like those currently plaguing Windows, and thus incapable of harming system files... but if someone exploits the backticks in
For school & work Linux systems I created a preconfigured freshrpms package which includes a cron job to regularly check the redhat errata, download any updated packages, and mail root when something new appears. It's a step in the right direction - Linux is a secure system because bugs are so quickly found and fixed, but it won't be publically perceived as a secure system if security-unconscious newbies never see or apply those fixes.
I suspect redistributing the Buffy episode may be illegal, and I'm certain it's pissing off the WB, but it's hardly immoral.
Don't get me wrong - I believe in intellectual property, I even believe license writers should be allowed to enforce any and every wacky fascist provision they can sucker a customer into agreeing to - but those are cases based on mutual agreement. If you don't like a software license, don't buy the software. If you think Hollywood is ripping you off with $7 movies, don't buy a ticket. If you're pissed that you can't photocopy or web publish a copyrighted book, write your own content. And if you distribute warez, whether it's commercial software or an mpeg of The Matrix, you are basically distributing stolen property.
I'm much less concerned about the morality of republishing a TV broadcast. There's no agreement, stated or implicit, made by someone whose only activity was to intercept radio waves streaming through their own homes and save a copy of the data. There is an exchange which occurs with every other distribution of copyrighted data (hell, even with cable & encrypted satellite broadcasts technically); an exchange that just doesn't occur with broadcast media. If WB wants to retain control over their TV programs, they should have asked us before sending those programs to millions of households.
Granted, it's currently used to refer to Unix sysadmins more than programmers, but there's a pretty big overlap between the two groups.
Let's all start making our emails seven-bit-clean with "red flag" armor - choose, say, 65536 words like "assassination" or "echelon" that are certain to trip email scanning bots, then map every 16 bits of an encrypted message to the red flag word at that index in the word database.
The irony of thousands of innocuous messages, both encrypted and tailored to fill NSA "suspicious message" databases, would be amusing.
If the frame rates on this thing are in the ballpark with the Voodoo 3 3000 (and last I heard they weren't much slower, even with 32 bit rendering on), I'm getting one. I'm worried about performance on my K6II, though - all the benchmarks I've seen show 3dfx cards performing almost evenly with Celerons at the same clock speed, while TNT/TNT2 performance falls off dramatically.
I know a couple other people whose only gripe with the TNT2 was the lack of Linux support - I expect their minds are made up solidly now.
From the archives at www.geek-girl.com
Linux kernel 2.2.x vulnerability/exploit
Piotr Wilkin (pwl@WOTAN.2SLO.WAW.PL)
Tue, 1 Jun 1999 17:43:17 +0200
Messages sorted by: [ date ][ thread ][ subject ][ author ]
Next message: Salvatore Sanfilippo -antirez-: "whois_raw.cgi problem"
Previous message: aleph1@UNDERGROUND.ORG: "New Allaire Security Bulletin (ASB99-09)"
I'm sorry if this has been noticed before, but since I did't find anything
in the archives, I post it here.
There seems to be a bug in kernels 2.2.x (tested on 2.2.7 and 2.2.9), that
causes them to panic when they are sent a large number of specific ICMP
packages. I think the problem comes from the combination of the mangled
header length (shorter or longer ihl's don't cause hangup) and the random
ICMP packets (random type/subtype and source address) this program sends.
Windows 9x and FreeBSD 3.0 seem to be unaffected.
I think the most interesting thing is the date, though... I'm sure I'm making a timezone mistake here, but isn't that 8 hours ago? Is that faster or slower than the Linux teardrop fix?
It's annoying to find out about a new DOS attack, but the resolution is all that you could hope for.
It's a little less annoying that there don't seem to be any outstanding instant-crash attacks against Win98 to laugh about - they finally fixed the series of attacks that crashed 95 for 8 months straight, and I haven't seen anything since. Did Microsoft finally get their IP stack right?
If you want to use files larger than 2 GB (the largest number that will fit in a signed 32-bit integer), then get a 64-bit system to put them on. If you want to have simple, efficient, easy to code and easy to port seeks within your files, then you're going to want to be able to use a signed integer to seek back and forth (let's not even mention the trouble with mmap() if your files are larger than a pointer can index...)
The *last* thing Linux should do about 2GB files is try and use hack after kludge to satisfy people who want to use Intel chips but don't want to hear about their limitations.
Because the second time I watched The Matrix, I caught the half-sentence that had escaped my (and apparantly everyone else's) attention earlier: (not an exact quote): "with this bioenergy, plus a cheap form of cold fusion..."
In other words, they're getting gigawatt after gigawatt out of ocean water; the "human battery" thing is just some twisted thing that evil computers do for kicks.
Of course, I would have LOVED it if they had thought of another poster's suggestion: that the apocalyptic world was entirely humanity's fault, and that the Matrix was a "Zeroth Law" reaction of the computers, meant to keep humanity happy and prevent them from hurting each other, for our own good.
It would have lost the "pure evil" factor, but it would have allowed a lot of cool complications to develop in a four-sided war:
The computers who want to help humanity
vs.
The humans who want freedom at any price
vs.
The humans who want to cooperate to produce virtual utopia
vs.
The computers who are sick of supporting inferior life and want to wipe out the whole project to make room for themselves.
But I don't want a Palm Pilot, I don't want a Rio, I don't want any of the one-trick gadgets on the market today.
What I want is a portable gadget with wireless 'net access, a built in cell phone (preferably embedded in the PPP stream so I don't have to log out to call someone), sound quality good enough to play MP3s, image quality good enough to run snes9x (even if only in black & white), CPU speed to do both, hardware open enough to run Linux on, a 10baseT port instead of some stupid serial-based "cradle" for uplinks, a set of long-life, built in rechargeable batteries, a fold-out keyboard, a stylus, and a couple PCMCIA slots. (an optional PCMCIA-sized GPS receiver or 4 GB drive would be nice too) Oh, and *lots* of RAM. At a dollar a meg, putting 32 or 64MB RAM into a handheld device isn't ridiculous. I don't know what default software should be included, but at the very least I should be able to look at the bundled math program and throw away my TI-85 (or install Linux/matlab and throw away my TI-85).
This would probably cost over a thousand dollars and require daily recharging with moderate use today, but it is possible, and those numbers are only getting rapidly better. I guarantee the first company to replace the PDA, pager, cell phone, game boy, walkman, calculator, GPS receiver, etc. with one single unit, and make it affordable, will be rich overnight.
What happened to the other problems with FTL travel?
I'm thinking specifically here about the more slippery definition of "simultaneous" in special relativity, and it's consequences. Assume a ship that can:
a. Accelerate to any sublight speed (and thus any "normal" reference frame) arbitrarily fast.
b. Use "Warp Drive" to move between two points in space time faster than light.
That's all you need, and assumption a. is perfectly in accordance with physical laws (and technically plausible with some externally powered propulsion system).
Throw in the cute fact that for any two points in spacetime that are not in each other's light cones, there is an inertial frame of reference where those two points are simultaneous, and your ship can:
1. Warp 1000 light years away, simultaneous with the frame of reference of the solar system.
2. Accelerate to near light speed.
3. Warp back to earth, but 999 years before it left.
I wish I could link in a diagram of this... but go look in a physics textbook; it's a classic paradox meant to show why FTL travel is impossible.
I've never heard a good explanation of how Alcubierre's theory (not to mention whatever new concept has come up) deals with this.
I'm surprised nobody seems to have mentioned this so far: With an emulator, I may be able to illegally play pirated games... But I am also able to legally play games which I have purchased, even though I don't own the console system they were designed for.
This is something Nintendo wants to stop, every bit as much as software piracy. It probably hurts their revenues just as much (how many people would pay $200 for a console system that only plays games the $2000 computer they own plays better?), and since it's a completely legitimate practice they can't attack it. I don't own an N64, but if I could get my hands on UltraHLE I'd be happy to borrow a friends Zelda64 cart, download the ROM, and play through it on my computer, deleting the ROM when I'm done. I bet it would look better on a Voodoo 2 than a low-res TV set, too.
The only thing Nintendo can do is change the subject - pretend that because emulators can be used to pirate games, they must only be used to pirate games. Claim that the data on a game cart is somehow an integral part of the cart, not software to be used as the legal owner sees fit. Even claim that their consoles aren't legally computers, and aren't subject to the same rules restricting software licenses.
If I transfer the contents of my hard drive to a faster model, then throw the old drive in the garage, I'm not violating any software licenses. If my old hard drive crashed irreparably and I get copies of my own software from a friend, again I'm doing nothing immoral.
And if I buy a N64 cart, then decide to play it on a quality 3d card/monitor rather than on an N64/grainy TV, even if I never buy an N64 in the first place, that's OK with me. I'm doing nothing wrong, the emulator authors didn't do anything wrong.
The only people in this scenario doing anything close to wrong is the N64 manufacturer attempting to boost sales by artificially maintaining a software lockin.
I thought they might be forming a group to do the only thing that might hurt Linux: make Windows NT efficient, rock stable, compatible with decades of portable code, truly multiuser, secure, and network transparent.
But naah, they're just going to spread some FUD.
Move along folks, nothing to see here.
Caldera Open Administration System, that is. The article gave it a passing mention.
/etc is called for. I'm hoping we'll see enough improvement with Linuxconf or COAS that Joe Avg. Macuser will be able to handle those tasks as well.
/etc/ld.so.conf "when adding new libraries" - does he just mean new libc5 libraries, or would I really have to futz around pointlessly every time I make a semi-weekly upgrade to some bleeding edge libs? If it's the latter, count me out. Red Hat's been supporting libc6 & libc5 programs side by side for a year and a half now without that kind of kludge, and all the trouble I've ever had to take was run "ldd" on libc5 programs to make sure all the libraries they want are in /usr/i486-linux-libc5/lib
It's been in quiet development for at least a year now, Linux Journal claims it's going to be modular, GUI based & vi-compatible, it looks like the only serious competitor to Linuxconf, and it fills the software hole (user friendly & newbie friendly system configuration) that Linux needs most desperately.
Linux is currently at a state where any PC or Mac user could switch to and use it, as long as they had some guru to log in as root whenever mucking around in
A couple more things I'd like to have cleared up:
Is COAS under the GPL like they said it would be? What non-free software (if any) is on the Caldera 2.2 CDs?
Petreley mentions having to muck about with
The best solution, of course, would be if Corel recompiled WP8 (or hurried up on their 2000 product) so I could ditch libc5 compatibility entirely.
I agree. Although I think it is strange that you have not mentioned any lisp dialect here. Guile had been perfect.
I'll agree in theory - although I personally am not a comp sci major and haven't been forced to wrap my brain around functional programming. The idea of say, summing a list via recursion instead of a loop just doesn't feel natural.
I also don't know enough about Guile specifically - how easy is it to interface with code written in other languages?
And BTW, you don't really mean that you prefer writing AI in C?
Just an example, just a theoretical example... but one point I was trying to make is that if you can link to C code, it's possible to write wrappers as a "gateway" to just about any language out there for people who really hate your particular scripting choice.
Yet Another Scripting Language, that is.
And not a good one either - what were they thinking? While SLIC does not directly support user defined functions, there is a workaround. I'm sorry; I'll get by without object oriented programming fine, but at least give us functions to work with. I can't even imagine writing a script longer than a couple hundred lines in this. Not to mention the undeclared variables (what do you mean there's a typo in my script? There was no parsing error!), and the lack of locally scoped variables, structures, pointers/references, or many other features of every useful language created in the last 30 years.
They would have had less programmer work, probably earned greater performance, and definitely made the game infinitely more extensible, if they'd added a Perl, Python, or even Tcl interpreter instead of creating this travesty. (Hell, Java, Javascript, etc. would work too if the licensing was OK.)
They also would have tapped in to the vast mountain of existing code out there. Imagine a python script that, at the end of every turn, updates a web page with a summary of the game. Imagine a perl script with the Gimp module, allowing you to make animated GIFs of the battles when you press a button. Imagine writing (and compiling) your own AI in C, then hooking it up to libtcl, perl XS, or whatever...
Not that this is all bad - I mean even crippled user extensible scripts are definitely preferable to no user extensibility at all... but they hit on an idea that could have been as revolutionary to strategy games as QuakeC was to 3D shooters, and it hurts to see they missed the target.
Most of this can be explaned as the use of MP3s for pretty much all of the sounds in the game. They're very processor-intensive little buggers.
That's actually not quite true - anything Pentium or better should have more than enough CPU cycles for decoding MP3s - the problem is that MP3 playing is both somewhat CPU intensive and real time, which means if your MP3 player doesn't have special priviledges and goes 10ms without running, it's going to skip.
Off topic: does anyone know which MP3 players for Linux can do a sched_setscheduler() when they're suid root? x11amp 0.6 did this, and on my system it made the difference between occasional skips and unskippable music - but x11amp 0.6 is getting old.
Anyway, Centauri should still have more than enough CPU cycles left over to animate a small sprite or two quickly, even if every sound effect was being decoded on the fly (and if they're not caching wav data of the small sound effects, they've got a coding problem right there). It honestly feels like the whole game is being run in an interpreter, like it's some Tcl/Tk or uncompiled BASIC app. That doesn't sound likely, but I can't think of anything else that might be sucking up that much CPU.
If they do get a Linux port out, I hope they take the time to fix their atrocious speed problems. Don't get me wrong, it's a great game... except that it was running slower on my friend's PII-400 than it should run on a P166. I'm sorry, there's no reason that popping up an icon menu or scrolling across a 2D map should be anything less than instantaneous on a modern computer.
The lack of a Linux port (when Civ:CTP has one in the works) was one strike against my buying Alpha Centauri, but the fact that Centauri would be unplayable on half my friends' and relatives' computers was just as bad.
Let me get something clear: I don't expect Microsoft to ever do anything more than say "You can order most of our Windows 2000 source code for $200 as long as you sign this agreement not to redistribute it or any patches to it except to us, not to work on any other operating system or Windows API related projects, and to give us your eternal soul." That kind of "new type of open source" would have little or no effect on Windows itself - except it would make life much easier for warez dudez who want to make trojan software and hackers looking for remote exploits.
What I'm worried about is that some clueless but well-meaning judge might decide that the best way to crack down on Microsoft would be to whip out the "eminent domain" stick and take away Windows by making it true open source. That would be OK for free software in general, but for the state of computing it would be a horrible step backward.
I don't just want free software, I want free, good software. We don't need hackers poring over Windows 2000 in some vain attempt to make remote display of apps stable and scalable, or trying to fix whatever hideous kludges they've got preventing them from throwing away DOS finally.
For those people who say that "we haven't seen the source code, how do we know it sucks", we've got enough good evidence to go on. Their repeated attempts and repeated failures to get rid of the 16 bit code and DOS core in Win9x is one example. The travesties of the Win16 API and WINS is another. The fact that they've had a half-baked 32-bit port of NT to the Alpha instead of writing 64-bit clean code in the first place is a third. How about the WinNuke/Teardrop/Syndrop/Newtear/etc series of attacks on the Windows IP code, where the patches took weeks or months to be released, and where half of the new attacks were just minor changes (e.g. using UDP instead of TCP) on the "patched" attacks?
I'm told that NT has some good kernel design decisions, and that it's just the wacky video drivers and the mess of the Win32 API weighing it down... but I'm told this by people who still claim that NT is a "microkernel" architecture.
I'm also personally impressed with OLE, which I think is one example of something Microsoft did themselves, did without too much prior art (All I can think of is ToolTalk in the Unix world - apologies given and corrections requested if I'm missing some Mac history here), and did well... but the underlying implementation isn't of value here (and is probably inferior to CORBA if they're just making it network transparent now), the API (which we already have and which Gnome is practically cloning AFAIK) is.
In short, it's evident that with the exception of accelerated Wine development, an open source Windows would be fairly valueless.
What an open source Windows would do is something I'd rather not see happen: insure the immortality of Windows. More than ensuring "World Domination", open source software is ensuring that free Unix will never die... and for both obvious and personal reasons I hope Windows does die. In the next 5-10 years, please. The idea of Windows dying may have seemed ridiculous a couple years ago, but with Linux pulling out the rug from under NT server and threatening the desktop, with the Godzilla movie of Windows 2000 fast receding, with 64 bit NT development headed by the guy responsibile for Windows 3.1 (which scarred me emotionally), and with even Apple clawing their way back out of the grave...
One of the best things about Unix is that, incompatible vendor extensions aside, all the Unices are built around a set of common, well-designed APIs that give the word "standard" real meaning. I'd hate to see Windows become a standard as well, not because of any design features but because people got to make a lot of cheap copies.
Forget about Linux distro's....
Every damn system is different!
Damn straight. I'm using bleeding-edge releases of my kernel, the C & C++ libraries and compilers, and a few other things at home that I won't be installing at work until they've been hammered on a while. My system here is using Window Maker, at work it's FVWM2, and my roommate is using KDE. He also drives a different car, wears different clothing, and likes different food. It's called "freedom of choice", and we like it.
For the record, all of the systems we own or run execute the exact same software binaries fine, (and the LinuxPPC systems run the same software code fine after a recompile).
We got libc5 libc6.
You forgot libc4. That's right, we've got old systems that only run libc5 binaries, ancient systems that only run libc4/a.out, new systems built around libc6, and new systems that run all three. We've got the same programs recompiling with few or no source code changes on the newest systems, but getting the technical benefits nonetheless. Read up on the technical differences between C libraries, and try to stretch your mind to encompass the concepts of "progress" and "backwards compatibility" at the same time. Once you've grokked that, email Microsoft and ask them why most of their drivers, DOS, and Win16 programs don't work with Windows "New Technology". Then ask them why most of their Win32 programs still don't work with "Windows 2000" without source code changes. Then ask them what's going to happen to that 32-bit specific API when they finally push 64-bit NT (a separate project from Windows 2000...) out the door three or four years from now.
Different widget sets, window managers,
Yeah, yeah: "Ein MFC, Ein Windows, Ein Microsoft." Great world for everyone to be forced to live in. Different widget sets and different window managers fill different needs of programmers and users, accomodate wider sets of preferences, and in general foster competition and evolution of software. Anyone horribly confused by a GTK program sitting on the same desktop as (or an adjacent virtual desktop to) a Lesstif program should probably just give up entirely and get WebTV.
hell even different windowing systems (MetroX,
XFree86 et al)
If you think those are different windowing systems, you need to go back and do some more reading. The exact same programs run on either and can't tell the difference. Most hardware works with either happily. In fact, chew on this: not only are different X servers not incompatible, they are so compatible that I can run decades old HP-UX programs on the same desktop next to new Linux software, with the only noticeable change being that Qt and GTK look better than Motif.
A lot of boxes run kernels so well tuned
you can't even boot it on another machine!
Yes, boxes whose owners have chosen to make those tuning changes themselves. Nobody's holding a gun to your head to force you to use 486 or Pentium II specific instructions, or to get rid of your EIDE drivers on a SCSI-only machine, you know.
I guess is too bad it all works, after all thats what really counts.
Try not to sound too disappointed.
I really don't understand the backlash here - for most people, if Linux isn't for you, don't worry about, ignore it in the papers and the trade rags, and it won't bother you again until they start advertising idiot proof distributions in 2005. If you're a Win9x user and happy that way then Linux just isn't a factor.
If you're an NT server programmer watching your world start to crumble, on the other hand, I can recommend several good books on the POSIX standard that you may find of interest in the near future.
Currently, the only license information I can find distributed with the kernel itself is the GPL, in /usr/src/linux/COPYING. However, there seems to be the exception, spoken by Linus himself, that binary kernel modules are allowed as long as they don't require interface changes in the kernel proper. We've already had one such case - /dev/3dfx
If this is to be the case, we had better get it in writing and get it in writing ASAP - I can just picture the chaos when binary-only drivers start becoming common, and some kernel developer says "No, I contributed to the linux kernel under the GPL. Stop allowing binary modules, or take my 50K lines of code out of the kernel."
Is that it takes three ideas (a pure OOP language, a portable virtual machine & byte code specification, and a nice GUI library) with lots of promise, then it locks those ideas irrevocably together. If developers could compile to native code, link to existing object files, etc. Java would have a lot broader usefulness. Good job, Cygnus.
The argument essentially is "If a hacker reinstalls your operating system, and they have the source code, they can plant trojan programs." This seems to be the computer equivalent of "If a murderer is hiding in your kitchen, and you own steak knives, they can stab you when you come to breakfast." Technically true, but missing the point entirely. If someone is trojaning your Unix, then they already have root, and you've already been cracked - it doesn't much matter what tools they use to make use of that crack.
He also argues that closed source makes this impossible because they can... er... put up really distinctive splash screens... no, wait, that's not enough... dialog boxes! They can put up distinctive dialog boxes so that the users will see the dialog boxes and know they haven't been hacked.
Would someone please send this guy a trojan Solaris login binary or a copy of Back Orifice to look at? Hell, with the obvious emotional investment and feelings of martyrdom he's showing, even a VB app that puts up the Windows login dialog might be enough to shatter him psychologically.
My personal preference would be to have (Threaded, Sort by Score) pages sort by the highest score in each thread, then display the scores for followups next to the followup names. This would have the disadvantage that the largest threads would invariably filter to the top, but it would mean the people who want to read all the Score:4+ posts and skim everything else (which is why we're sorting by score anyway) would be able to do so most easily.