I dunno, it's almost my main browser now that the PowerPC port is up. Waiting on M13 so I can have the PowerPc work in a milestone build, but I post with it most of the time.
I would think it might be, espescially if you disabled the debug stuff (./configure option). It would probably improve even more if you used pgcc or some such (though you might not get a working version, as pgcc does have some bugs in it's optimization. It's gonna take a long time to build it on a P100 though:-). On my powerPC G3 system, a full build takes like 1 gig of scratch space for/usr/src and about 2-2 1/2 hours. Not that this is not abnormal or bloat - similar sized packages take similar space to build, and the final executable isn't horrible. The kernel goes from 15megs source (tarball) to ~100megs objects to 2 megs of binary, and mozilla is just comparable.
You're going to have to tell us which milestone it was, since there haven't been any alphas or betas yet:-)
Personally, I really like the new chrome (think skin) in this M12-M13 CVS build (that I'm using to post this). And I'm sure that there will be others as soon as someone decides to do them (A good knowledge of CSS and the DOM would be required, but I'm pretty sure the whole thing is driven through HTML Style Sheets). The only thing that's annoying me right now is that the keyboard response is *really* slow - but that has a bug # and is release critical for M13. It looks small, somebody messed up the event priority (?I could be mis-reading, don't know that section of the lizard).
I also have service at my house (at school, I use the university network). Latency is reasonable, though not phenomenal - maybe 150 or so - and I regularly get 200+ k/sec. All for $30 a month. Perfect it may not be, but compared to my phone-lines (which are noisy anyway, so I never got past 28.8) it's wonderful.
This is more like me looking at my caller ID and not picking up, because it's the pay phone where I keep getting pranked from. I could pick up (a site can not follow any UDP they choose not to), but I don't want to.
Besides, only @home can really tell who the spammer is. see My other comment
OK, this is how it works. All that a recipient of spam can reliably know is the originating ISP (this is know, because the IP of the remote is known in ths logs and must be correct - this IP would be news.isp.net, probably). If this IP were to be incorrect (spoofing) then no data could have returned to the remote and the protocol would have failed, so no spam in the first place.
OK, so we do know ths ISP. However, the spammer can set the user account to absolutely anything. Therefore... One goes to the ISP, and nicely asks them to check this out. They would have logs of who was on when, and could probably figure out who actually did it (to them, you see, the remote would be the spammers own machine. They know who they gave which IP to, and can figure out the identity of the spammer). If they didn't keep the logs, they could watch for him to resurface (ie keep logs temporarily and deal with it if he does it again). This would almost certainly be acceptable as a response, and would get them off the hook.
If the ISP is uncooperative, then all that can be done (from the next level up) is to blackhole the ISP. Only the ISP can really know who the spammer was, so theu have to deal with it, or ignore it. If they ignore it, I guess the people at the next level have the right to recommend blocking and to block it for themselves. If you read the post, individual news-carrying sites have the choice if a) whether or not to follow UDPs at all or b) to follow them, but exempt this one. How to do so was described. So if you like spam or think this is too extreme (maybe it is, I haven't hear whan @Home's response was or how many warnings they got) find a news-carrier who is not going to follow it. Or run nntpd yourself, and don't follow it.
So it's a lot like the system right here at slashdot. Read at -1 if you want, or don't if you can't stand trolls and spam. Individual choice (except that you do have to be the person running the newsserver and using up the resources to be able to make the choice.
At least, that's my understanding of the setup, but I don't actually run a usenet server, so any innaccuricies are hereby disclaimed. If I'm wrong, would someone reply and say so.
Oh, by the way - Mac on Linux (which does require real PPC hardware, but, to my knowledge, not Mac hardware) can now use the 'New World ROM' image that is a file in the system folder of Macs. This file can also easily be extracted from the freely available ROM-patches for the iMac and G4, or copied from the System CD (8.5+?), which you would presumably want to have. An emulated Mac won't get far without MacOS.
Whether or not it is legal to obtain the ROMs in this way instead of by romdumping a sacrificial Macdo is left as an excersise to someone who is a lawyer. My Linux box *is* a Mac G3, so I feel legit.
The port is complete and maintained already. It's just not available. OS 8 was originally ported (I think), and, in any case, PPC is the 'young' port of OS X, since it's part BSD part Nextstep.
You're confusing the Rage Pro (which was not much:-) and the Rage 128. There aren't any iMacs with Rage 128, only Rage Pro (and a few with Rage II+). THe Rage Pro is not much of a card, I admit. But from what I've seen of the 128 (in person), it's decent, at least. Not phenomenal, but if you really want the trimmings (TV in/out,2d, reasonable 3d), it's not a bad deal. It's the Pro where Carmack was squezing blood from rocks (to quote you).
Because it's small, and they're not. Now if it was pretty too, that would be awesome. But it looks functional, and that goes a really long way in my book. Gnome and KDE both have lousy filers (sorry, they just 'feel wrong'). GMC wasn't stable and I never could get it to recognize apps it didn't want to. KDE puts those dang 'Templates' and 'Autostart' folders on my desktop (Maybe KDE2 will stop this? I could always buy more RAM then:-). Besides, both hate running without their window-manager/panel/sound-server/take-over-all-my -RAM companions. Maybe they've gotten better, but I retreated into Eterm, blackbox, and LyX, waiting for something I liked to arrive...
This is what the shared memory extensions are for! A client that want's more direct access (and already knows that there is a shared memory space) will use shared pixmaps, and poof, no sockets! The program puts data in RAM, the Xserver works from there. A program that wants even more direct access (and knows it's on the same machine) can use DGA to write directly into VRAM.
There is the main protocol (high abstraction) and two other ways, varying speed, transparency, and portability. Pick what you need.
Also, saying Ethernet is completely wrong. loopback TCP would be the worst case (unless there really is a network layer) and UNIX domain sockets should be what is being used.
Unfortunately, I don't do much in Xlib, so YMMV for details, but I think I'm close
I can see this as a really good idea - but the problem is the number of packages that have been developed (under GPL, with BSD the need to make a license change isn't a problem) by so many developers that no one has a clear title to the code to make the needed license changes to allow something like this. So only new, clean-roomprojects (or ones that aren't GPL'ed, or that require patch copyrights to be turned over to the 'main author' could benefit. Probably still worth pursuing
Re:This moderator is an idiot.. THIS is interestin
on
Laptop Pentium IIIs
·
· Score: 1
He's suggesting lowering the clock rate incrementally until the CPU is at 100% utilitation (slow it down until it can just barely keep up). This would save battery, but it would be crucial that the scheduler be able to bring the clock back up *fast* in response to a burst of activity, so that the computer doesn't feel sluggish. Some CPU's (powerbooks, this new one, etc) actually do change core clock frequencies on the fly.
It was the boot-floppies that weren't ready. And also some of the new ports that were supposed to be in potato (Debian/PPC, don't know what all else) weren't quite there.
Well, the $35 isn't nearly as much a big deal as the fact that MSFT itself is donating to the WINE project. Now that's funny.
That's what I had in mind when I said that. Or maybe they could help with what's really needed - specs for some core dll, or something else. Surely there must be some programmer at Micros~1 that could help out usefully.
Have you ever heard those stories where some wealthy businessman loans $1000 to a struggling student (or otherwise) and says "Pay it back when/if you can". Then, when the person comes back years later, he just says to repay him by doing the same for somwone else?
I think he ought to ask Microsoft to 'repay' him by helping some other company in need:-).
2.1.24 was the 'stable' reference for PowerPC right up to the 2.2.0 release. It was parallel-developed and patched, fixed, etc, (just as 2.0 was while 2.1 was being worked out) but 2.0 was not usable on PowerPC.
So, they probably did their development of the box during the pre-2.2 timeframe and used the kernel tagged as 'stable' by the LinuxPPC team at that time.
talk user@host:-) But, this would require that you know their hostname, I admit. Unless someone used one of the proxying talk daemons as a central setup.
It uses no linux utilities, because none exist. Linux is the kernel only, userspace is (mostly) GNU, with the occasional BSD util (few though). So correctly, it is GNU/Linux, and this will be GNU/Hurd. As aI know, though, only Debian gets this naming even moderately correct, though.
Slashdot Mirror
I dunno, it's almost my main browser now that the PowerPC port is up. Waiting on M13 so I can have the PowerPc work in a milestone build, but I post with it most of the time.
I would think it might be, espescially if you disabled the debug stuff (./configure option). It would probably improve even more if you used pgcc or some such (though you might not get a working version, as pgcc does have some bugs in it's optimization. It's gonna take a long time to build it on a P100 though :-). On my powerPC G3 system, a full build takes like 1 gig of scratch space for /usr/src and about 2-2 1/2 hours. Not that this is not abnormal or bloat - similar sized packages take similar space to build, and the final executable isn't horrible. The kernel goes from 15megs source (tarball) to ~100megs objects to 2 megs of binary, and mozilla is just comparable.
Personally, I really like the new chrome (think skin) in this M12-M13 CVS build (that I'm using to post this). And I'm sure that there will be others as soon as someone decides to do them (A good knowledge of CSS and the DOM would be required, but I'm pretty sure the whole thing is driven through HTML Style Sheets). The only thing that's annoying me right now is that the keyboard response is *really* slow - but that has a bug # and is release critical for M13. It looks small, somebody messed up the event priority (?I could be mis-reading, don't know that section of the lizard).
Deja it is.
Besides, only @home can really tell who the spammer is. see My other comment
I don't think so - BSD never gave me that hiccup deja-vu thing where the mouse freezes, then jumps. Only RedHat ever did that.
OK, so we do know ths ISP. However, the spammer can set the user account to absolutely anything. Therefore... One goes to the ISP, and nicely asks them to check this out. They would have logs of who was on when, and could probably figure out who actually did it (to them, you see, the remote would be the spammers own machine. They know who they gave which IP to, and can figure out the identity of the spammer). If they didn't keep the logs, they could watch for him to resurface (ie keep logs temporarily and deal with it if he does it again). This would almost certainly be acceptable as a response, and would get them off the hook.
If the ISP is uncooperative, then all that can be done (from the next level up) is to blackhole the ISP. Only the ISP can really know who the spammer was, so theu have to deal with it, or ignore it. If they ignore it, I guess the people at the next level have the right to recommend blocking and to block it for themselves. If you read the post, individual news-carrying sites have the choice if a) whether or not to follow UDPs at all or b) to follow them, but exempt this one. How to do so was described. So if you like spam or think this is too extreme (maybe it is, I haven't hear whan @Home's response was or how many warnings they got) find a news-carrier who is not going to follow it. Or run nntpd yourself, and don't follow it.
So it's a lot like the system right here at slashdot. Read at -1 if you want, or don't if you can't stand trolls and spam. Individual choice (except that you do have to be the person running the newsserver and using up the resources to be able to make the choice.
At least, that's my understanding of the setup, but I don't actually run a usenet server, so any innaccuricies are hereby disclaimed. If I'm wrong, would someone reply and say so.
The ones who do this can :-)
Whether or not it is legal to obtain the ROMs in this way instead of by romdumping a sacrificial Macdo is left as an excersise to someone who is a lawyer. My Linux box *is* a Mac G3, so I feel legit.
The port is complete and maintained already. It's just not available. OS 8 was originally ported (I think), and, in any case, PPC is the 'young' port of OS X, since it's part BSD part Nextstep.
Damn, I have a Rage II+. POS.
There is the main protocol (high abstraction) and two other ways, varying speed, transparency, and portability. Pick what you need.
Also, saying Ethernet is completely wrong. loopback TCP would be the worst case (unless there really is a network layer) and UNIX domain sockets should be what is being used.
Unfortunately, I don't do much in Xlib, so YMMV for details, but I think I'm close
It was the boot-floppies that weren't ready. And also some of the new ports that were supposed to be in potato (Debian/PPC, don't know what all else) weren't quite there.
Just got that far in the book "That man who loved only numbers". You're right :-)
I think he ought to ask Microsoft to 'repay' him by helping some other company in need :-).
P.S. Are you listening, WINE?
So, they probably did their development of the box during the pre-2.2 timeframe and used the kernel tagged as 'stable' by the LinuxPPC team at that time.
talk user@host :-) But, this would require that you know their hostname, I admit. Unless someone used one of the proxying talk daemons as a central setup.
Get yerself a copy of alien, turn rpms, debs, slackware packages, and binary tarballs into each other at will. Nice.
Cool, didn't know you could keep them from loading... I may have to try this.