While I don't condone weapons research, I think this is certainly interesting. If the RPGs flaunted around today were capable of Tomakawk-size destruction, i think we'd simple see skirmishes ending faster, in a "disease-burnout" kind of way. I'd hate to see this effect be used as weaponry by anyone, but if people are going to fight, the faster its over the better, in my mind. Maybe I'm mistaken?
Doctor Richard Gatling invented the gun that bears his namesake, and believed it would wreak so much carnage, war would become unthinkable. Well he batted.500...
> Also, aren't the NUMA and RCU multi-processor patents owned by IBM?
Doesn't matter, the suit is about copyright. If SCO licensed the patent, then produced code with IBM, Novell, cobbler elves, and the Trilateral Commission... anyway, if they own the copyright to the source code of an *implementation* of the patent, and that copyright is infringed, they still have a case. In fact, if the code is indeed copied verbatim, then someone did a bad somewhere, and the code need to be removed.
SCO of course doesn't want the code removed, as their own statements have made clear. This has bad faith and estoppel by laches written all over it... IANAL, but IBM seems to agree.
> the smart people downloaded the 1.2MB patch last month and had no idea anything was going on until we read about the worm on Slashdot.
The smart people who never run interactive sessions on their workstation as administrator, and are therefore unable to use windows update unless they remember to log in as administrator every now and then?
Yeah, those. I've been forgetting. Cripes, I must have a hundred megs of "critical updates" to download over my dialup now.
I can see hiding DTC, I can see hiding the Fax subsystem... But Pinball???
Nice.sig btw: Jane Roe: Roe v. Wade was a fraud
Here's another one from her on clinic bombings: "I personally think it's the pro-abortion people who are doing this to collect on their insurance, so they can go out and build bigger and better killing centers."
> Frankly, I'm surprised at all the negative reaction to the Segway
It was hyped. Boy was it hyped. As a super-duper-tippy-top-sekrit project, then this revolution in transit is unveiled...
And it's a scooter. It's a neat scooter, it's a high-tech next generation scooter. But it's a scooter all the same.
Forgive us for being underwhelmed, and directing derision and scorn at this pompous windbag who touted it as the most amazing invention since the automobile.
While we're on the subject, anyone have brand recommendations for a good cheap USB drive (HD or flash is fine). Preferably cheap, it doesn't have to be big at all. I just need it for putty and a few scripts.
> How do I make a window always-on-top in any version of Windows? No way that I know of unless the application supports it.
Did you try googling for "windows" "always on top"? There's oodles of little utilities, some in the 10's of K sizewise, that can nail a window's z-order (including "always on bottom" with some, which is really neat for monitoring meters and the like). Most are scriptable with the commandline, you just give it a window or application name.
I'd be interested in learning why more companies don't take a look into BSD environments. The security is there. The license is TOTALLY unrestrictive. It's stable, secure, well documented and well accepted (except on/.) -- why doesn't it get more corporate love?
Because they just use it. I'd say most of the firewall appliances out there run some form of BSD and not Linux, for example. It's just invisible. The BSD people are out to produce a good OS, so such "successes" are simply business as usual and not some domino in the chain leading to Eventual World Domination.
BSD continues on whether or not a bunch of advocates claim it's dead. It just does so in more invisible ways than you know.
Re:Former perl, python, java geek gone to Ruby
on
Ruby 1.8.0 Released
·
· Score: 2, Interesting
It's kind of funny- McCarthy didn't intend Lisp to keep the ultra-simple paren-based syntax, but to add an algol-ish syntax down the line.
Initially. Once macros came around, there was nothing like it in programming-land, and when you notice how many program structures you can implement with macros, there's still nothing quite like it, except perhaps for stuff like OpenC++ and camlp4, both of which have their *own* nasty syntax that doesn't look like anything else... Sexps became a feature very early on.
That said tho, it is a bit of a shame that there wasn't also a line-based alternate structuring convention and operator set that could elide most of the unnecessary parens. Good enough for haskell (layout rule and $ operator) after all.
> Is it just me, or does this kind of talk disturb anyone ?
Not quite as much as the spew from reactionary troglodytes that followed this post. Only to chip in my two cents though, yes it disturbs me a little. Killing's disturbing business, and being flip about it is on the low end of palatability. My guess is this guy got dressed down for saying it that way too...
What I do find more disturbing though is when the generals use terms like "objective neutralization", and everyone's favorite, "collateral damage". If we're killing people, then I'm all for the disturbing terms that remind people of that fact.
> Python is definitely not typeless. It's actually strongly typed. But it uses dynamic typing.
I consider allowing this behavior to be rather untypeful:
a = 0
foo(a)
a = "bar"
foo(a)
Changing the type of a, passing it to the same function, for which incidentally I can't overload on the type of its parameter... The fact that I can't add "foo" + 1 is only part of the picture. There's no place I can declare "a is an int, thou shalt not accept setting it to an incompatible type that thou canst not prove at compile time". I like loose typing for most scripting tasks, but there's places where I do want that discipline (especially when it's classes, not primitives like int). Yes, I thank &deity; that neither python nor ruby have perl's idiotic scalars, but it still lacks the typing I want, either explicit or inferred. Visual BASIC, of all things, has optional typing. How about it for python?
If you want to see an interesting container implementation (based on a trie, behaves like a hashtable) check out Judy Arrays. Judy is bogglingly complex, but it boils down to optimizing for the CPU cache. Judy looks fairly promising for some applications I'm looking at. The benchmarks I ran are sort of bogus though since I'm not causing fragmentation by doing other things between inserts and deletes. I suppose the moral of the story is, preallocate if you know your structure is going to get big.
> Why, for the love of God, would you want them under/usr/lib/cron, of all places?!
at was (still is sometimes) implemented in terms of cron. It made sense to keep cron's stuff together. I agree, it's an idiotic place, but there should be symlinks.
Putting them directly in/etc and not/etc/cron, incidentally, sucks. Directories are nice things, let's use them. Aside from keeping things nice and pretty, it would allow using some sort of remotable config database mounted as a directory, without having to move all of/etc to such a scheme or having to pull some sort of weird union mount (not available on Linux) trick. This of course assuming that lightweight usermode filesystems could be a reality on Linux...
But they're bringing in a "new World of Darkness" in August 2004. Maybe it'll be like Marvel's Ulitmates: a reimagining of the same concepts. Maybe there'll be compeltely different critters. Maybe it's something no one else has thought of. Who knows?
Yeah, but whatever they release, it's going to have copy that reads like this (verbatim from the WW website for Hunter)
For centuries, supernatural powers have reigned, warring among themselves, culling the human herds and lashing out from the shadows. The creatures of the night have held court since before the dawn of history.
Nothing lasts forever.
From the throng of humanity, individuals emerge who know the truth. They know monsters exist. Enough is enough. The forces of darkness must pay their due. You are the reckoning!
I remember being sort of impressed (not wowed, but pleasantly surprised) by the literate and articulate feel of V:tM. I'll grant that the writing quality has always been uneven, but garbage like the above as cover text is indicative of a very long fall, and like the Ragnarok of the Edda, I think most people are just going to write off the WoD's history without regard to a future installment. Don't know if they'll claim it'll end in fire or ice, but my guess is that it drowns in a sea of wretchedly poor writing like the above.
It basically reads (with many public statements to back it up): SCO is a failure. A miserable, abject, wretched failure. No one wants to buy our products. Our product has failed to meet customer expectations. To sum it up, we are dirt-eating belly-crawling failures. And it's IBM and Linux's fault.
Their stock is going UP on this report? Maybe it's stockholders thinking that SCO can win and that they'll get a piece of that pie when they file a shareholder lawsuit against the board for their failure (there's that word again) to uphold their feduciary duty.
Guess what? If I am a large multinational that makes, say, CRM software, and along comes some Free and kickass version that my customers start using then I start fucking laying people off or shipping my work overseas.
The day you can't pay people to create software better than other people do for free is the day you've fallen so far behind you can't even hit the baseline now. There's a few areas where labors of love (and the occasional outside company) are nearly category-killers: Apache, Linux, Mozilla and JBoss come to mind. But for those there's still Zeus, QNX, Opera, and Resin (all of which interestingly are sold on being leaner and faster). But if people stopped buying Photoshop because the Gimp did it all for them, it would be time to put Photoshop out to pasture.
This isn't to say free software isn't quality software, or even that it brings up the rear -- it's to say that if you can't sell something people are willing to pay for, you weren't likely to succeed much when the first real competitor came around, to say nothing of Free Software.
> How do you propose that people download alternative browsers if IE is totally removed from Windows?
FTP perchance? Besides, even if some sort of freak cosmic harmonic convergence happened, a court made MS remove IE, and MS complied, I imagine that would mean removing IEXPLORE.EXE and most of SHDOCVW.DLL, but not URLMON.DLL... so you could still grab a file with explorer by typing in the full URL, or using a shortcut.
I just don't fathom the "if" being a reality tho. But if it were, I rather imagine that the box stores *would* be selling browsers in boxes (or more likely OEM's would have it preinstalled)
> And another thing, who on earth are the Objectivist Center and Reason Foundation??
Followers of Ayn Rand (affectionately known in some circles as "Randroids"). You gotta read Atlas Shrugged, or at least The Fountainhead to get into Rand's philosophy known as Objectivism, which is not without its merits, but is predicated on an interesting mixture of shoddy logic (it loves tautologies) and vitriol toward fictional strawmen constructions of opponents (anyone who doesn't believe in absolutely unregulated capitalism) that makes the Two Minute Hate look like a love-in. I'm not going to pan it completely, it's a decent branch of existentialism, but it's not well-known for being very self-critical or indeed taking criticism of any sort gracefully. More of a culture problem than a belief one.
Again, her distinctly populist philosophy is called Objectivism, and thus you get names like The Objectivist Center. TOC (nee IOS, or Institute for Objectivism Studies) is actually one of the more moderate groups, because Rand's "official" legacy is carried on by a pinhead by the name of Leonard Piekoff of the Ayn Rand Institute who, to put it charitably, is nuttier than a fucking crate of baklava. Strident and dogmatic doesn't begin to describe Piekoff... but those internal politics are another story.
opposing viewpoints aren't terribly well-organized, but my favorite is a paper called The Unlikeliest Cult In History. I suggest reading it only after reading one of the books (I'm sure you can do a very uncapitalist thing and download a copy over p2p), since you won't really know where it's coming from otherwise.
A broader look at the subject would show a parallel with a more modern topic: anti-depression medications. There are plenty of examples of highly creative people -- geniuses in their fields -- whose creativity would likely have been quashed if they'd had access to a good Selective Seratonin Reuptake Inhibitor. Poet Emily Dickinson [washingtonpost.com] and artist Vincent Van Gogh [vangoghgallery.com] come to mind, but I'm sure there are many others.
I'm so glad Virginia Woolf and Van Gogh were suicidal, otherwise they just wouldn't have been any use to me in creating all those great works. They died at just the right time too, none of that mediocrity that might have resulted from happiness should they have found it.
For every one of your van goghs, there's thousands of geniuses with creative potentials who spend their days in a miasma of lethargy and isolation, creating nothing, because of depression. Maybe it's just the bipolar folks who are the most entertaining...
I'm no fan of the overmedication going on, but I think you better start taking a critical eye toward your own judgements.
> Laser beams and ion canons only to get pulverized because they're not defended against a good ol' fashioned shelling.
Methinks a good megawatt laser would do a pretty decent job blowing up said incoming shells, no? My guess is they use the beam weapon to replace those old phalanx point defense guns, which do a decent job against missiles, but really have a pitifully short range.
The RPM/DEb ideas are really good. The main problem i have however , and don't really know how can be solved is combining binary packages with source code packages (eg. when you compile you own X).
The FreeBSD ports system solved this a long time ago. Its dependency manager is based on make. It has a library of commands in sub-makefiles to check for things like libraries being installed, so for example when you install an app that requires qt, it checks for libqt.so using commands like ldconfig. It doesn't check for the package, it simply checks for the actual files that it depends on, and if they're not present, it builds and installs the port that contains the dependency. And ports are installed as packages and can be managed with the package manager. You get the best of both worlds. What's great about ports is that if you want to tweak something in the makefile, like options, or in the source, you just go into the source subdir, do it, then go back to the port and "make reinstall". You get your package built just the way you want.
RPM is capable of file-based dependencies instead of package-based, but the way RPM's have been managed, they typically don't (or they get it woefully wrong, as I've seen with RPM's looking for perl in a specific and wrong place, and not even trying a "which perl"). Thus you get the web of dependencies problem with RPM. debs are similar, but dpkg and apt manage them so well, there's rarely a real problem. There's a facility to build "fake".debs, but it's pretty cumbersome to do.
When the time comes and you want to update , let say Libc, then you will be unable to do so because the dependecies include almost every package.
Library versioning is typically resolved by simply keeping the old version around and installing the new package side-by-side with the old one.
Slashdot Mirror
While I don't condone weapons research, I think this is certainly interesting. If the RPGs flaunted around today were capable of Tomakawk-size destruction, i think we'd simple see skirmishes ending faster, in a "disease-burnout" kind of way. I'd hate to see this effect be used as weaponry by anyone, but if people are going to fight, the faster its over the better, in my mind. Maybe I'm mistaken?
.500...
Doctor Richard Gatling invented the gun that bears his namesake, and believed it would wreak so much carnage, war would become unthinkable. Well he batted
> Also, aren't the NUMA and RCU multi-processor patents owned by IBM?
... anyway, if they own the copyright to the source code of an *implementation* of the patent, and that copyright is infringed, they still have a case. In fact, if the code is indeed copied verbatim, then someone did a bad somewhere, and the code need to be removed.
Doesn't matter, the suit is about copyright. If SCO licensed the patent, then produced code with IBM, Novell, cobbler elves, and the Trilateral Commission
SCO of course doesn't want the code removed, as their own statements have made clear. This has bad faith and estoppel by laches written all over it... IANAL, but IBM seems to agree.
> the smart people downloaded the 1.2MB patch last month and had no idea anything was going on until we read about the worm on Slashdot.
The smart people who never run interactive sessions on their workstation as administrator, and are therefore unable to use windows update unless they remember to log in as administrator every now and then?
Yeah, those. I've been forgetting. Cripes, I must have a hundred megs of "critical updates" to download over my dialup now.
it's a hoot to see how much damage you can do when you're standing in a tower of 3 dozen Master Chiefs and your assault rifle fires frag grenades.
Am I the only one that read that as "a tower of 3 dozen Master Chefs" the first time? Oh the humanity! Bouillabaisse everywhere!
This line just kills me:
.sig btw:
Pinball=ocgen.dll,OcEntry,pinball.inf,HIDE,7
I can see hiding DTC, I can see hiding the Fax subsystem... But Pinball???
Nice
Jane Roe: Roe v. Wade was a fraud
Here's another one from her on clinic bombings: "I personally think it's the pro-abortion people who are doing this to collect on their insurance, so they can go out and build bigger and better killing centers."
> Frankly, I'm surprised at all the negative reaction to the Segway
It was hyped. Boy was it hyped. As a super-duper-tippy-top-sekrit project, then this revolution in transit is unveiled...
And it's a scooter. It's a neat scooter, it's a high-tech next generation scooter. But it's a scooter all the same.
Forgive us for being underwhelmed, and directing derision and scorn at this pompous windbag who touted it as the most amazing invention since the automobile.
Are you starting to get the picture?
While we're on the subject, anyone have brand recommendations for a good cheap USB drive (HD or flash is fine). Preferably cheap, it doesn't have to be big at all. I just need it for putty and a few scripts.
Or are they all pretty much the same?
> How do I make a window always-on-top in any version of Windows? No way that I know of unless the application supports it.
Did you try googling for "windows" "always on top"? There's oodles of little utilities, some in the 10's of K sizewise, that can nail a window's z-order (including "always on bottom" with some, which is really neat for monitoring meters and the like). Most are scriptable with the commandline, you just give it a window or application name.
I'd be interested in learning why more companies don't take a look into BSD environments. The security is there. The license is TOTALLY unrestrictive. It's stable, secure, well documented and well accepted (except on /.) -- why doesn't it get more corporate love?
Because they just use it. I'd say most of the firewall appliances out there run some form of BSD and not Linux, for example. It's just invisible. The BSD people are out to produce a good OS, so such "successes" are simply business as usual and not some domino in the chain leading to Eventual World Domination.
BSD continues on whether or not a bunch of advocates claim it's dead. It just does so in more invisible ways than you know.
It's kind of funny- McCarthy didn't intend Lisp to keep the ultra-simple paren-based syntax, but to add an algol-ish syntax down the line.
Initially. Once macros came around, there was nothing like it in programming-land, and when you notice how many program structures you can implement with macros, there's still nothing quite like it, except perhaps for stuff like OpenC++ and camlp4, both of which have their *own* nasty syntax that doesn't look like anything else... Sexps became a feature very early on.
That said tho, it is a bit of a shame that there wasn't also a line-based alternate structuring convention and operator set that could elide most of the unnecessary parens. Good enough for haskell (layout rule and $ operator) after all.
> Is it just me, or does this kind of talk disturb anyone ?
Not quite as much as the spew from reactionary troglodytes that followed this post. Only to chip in my two cents though, yes it disturbs me a little. Killing's disturbing business, and being flip about it is on the low end of palatability. My guess is this guy got dressed down for saying it that way too...
What I do find more disturbing though is when the generals use terms like "objective neutralization", and everyone's favorite, "collateral damage". If we're killing people, then I'm all for the disturbing terms that remind people of that fact.
> You left-wing whiners sure like to cut history's greatest mass-murderer a lot of slack, don't you?
Strange.. I don't see a lot of Bush bumper stickers alongside the "Free Tibet" ones. Did I miss a meeting recently?
> Python is definitely not typeless. It's actually strongly typed. But it uses dynamic typing.
I consider allowing this behavior to be rather untypeful:
a = 0
foo(a)
a = "bar"
foo(a)
Changing the type of a, passing it to the same function, for which incidentally I can't overload on the type of its parameter... The fact that I can't add "foo" + 1 is only part of the picture. There's no place I can declare "a is an int, thou shalt not accept setting it to an incompatible type that thou canst not prove at compile time". I like loose typing for most scripting tasks, but there's places where I do want that discipline (especially when it's classes, not primitives like int). Yes, I thank &deity; that neither python nor ruby have perl's idiotic scalars, but it still lacks the typing I want, either explicit or inferred. Visual BASIC, of all things, has optional typing. How about it for python?
If you want to see an interesting container implementation (based on a trie, behaves like a hashtable) check out Judy Arrays. Judy is bogglingly complex, but it boils down to optimizing for the CPU cache.
Judy looks fairly promising for some applications I'm looking at. The benchmarks I ran are sort of bogus though since I'm not causing fragmentation by doing other things between inserts and deletes. I suppose the moral of the story is, preallocate if you know your structure is going to get big.
> Why, for the love of God, would you want them under /usr/lib/cron, of all places?!
/etc and not /etc/cron, incidentally, sucks. Directories are nice things, let's use them. Aside from keeping things nice and pretty, it would allow using some sort of remotable config database mounted as a directory, without having to move all of /etc to such a scheme or having to pull some sort of weird union mount (not available on Linux) trick. This of course assuming that lightweight usermode filesystems could be a reality on Linux...
at was (still is sometimes) implemented in terms of cron. It made sense to keep cron's stuff together. I agree, it's an idiotic place, but there should be symlinks.
Putting them directly in
Yeah, but whatever they release, it's going to have copy that reads like this (verbatim from the WW website for Hunter)
I remember being sort of impressed (not wowed, but pleasantly surprised) by the literate and articulate feel of V:tM. I'll grant that the writing quality has always been uneven, but garbage like the above as cover text is indicative of a very long fall, and like the Ragnarok of the Edda, I think most people are just going to write off the WoD's history without regard to a future installment. Don't know if they'll claim it'll end in fire or ice, but my guess is that it drowns in a sea of wretchedly poor writing like the above.
It basically reads (with many public statements to back it up): SCO is a failure. A miserable, abject, wretched failure. No one wants to buy our products. Our product has failed to meet customer expectations. To sum it up, we are dirt-eating belly-crawling failures. And it's IBM and Linux's fault.
Their stock is going UP on this report? Maybe it's stockholders thinking that SCO can win and that they'll get a piece of that pie when they file a shareholder lawsuit against the board for their failure (there's that word again) to uphold their feduciary duty.
Guess what? If I am a large multinational that makes, say, CRM software, and along comes some Free and kickass version that my customers start using then I start fucking laying people off or shipping my work overseas.
The day you can't pay people to create software better than other people do for free is the day you've fallen so far behind you can't even hit the baseline now. There's a few areas where labors of love (and the occasional outside company) are nearly category-killers: Apache, Linux, Mozilla and JBoss come to mind. But for those there's still Zeus, QNX, Opera, and Resin (all of which interestingly are sold on being leaner and faster). But if people stopped buying Photoshop because the Gimp did it all for them, it would be time to put Photoshop out to pasture.
This isn't to say free software isn't quality software, or even that it brings up the rear -- it's to say that if you can't sell something people are willing to pay for, you weren't likely to succeed much when the first real competitor came around, to say nothing of Free Software.
You have dynamic IP addresses in there. Fuckwit.
> How do you propose that people download alternative browsers if IE is totally removed from Windows?
FTP perchance? Besides, even if some sort of freak cosmic harmonic convergence happened, a court made MS remove IE, and MS complied, I imagine that would mean removing IEXPLORE.EXE and most of SHDOCVW.DLL, but not URLMON.DLL... so you could still grab a file with explorer by typing in the full URL, or using a shortcut.
I just don't fathom the "if" being a reality tho. But if it were, I rather imagine that the box stores *would* be selling browsers in boxes (or more likely OEM's would have it preinstalled)
Pardon, that last link should be:
The Objectivist Center. Really should verify those links before typing 'em in...
> And another thing, who on earth are the Objectivist Center and Reason Foundation??
Followers of Ayn Rand (affectionately known in some circles as "Randroids"). You gotta read Atlas Shrugged, or at least The Fountainhead to get into Rand's philosophy known as Objectivism, which is not without its merits, but is predicated on an interesting mixture of shoddy logic (it loves tautologies) and vitriol toward fictional strawmen constructions of opponents (anyone who doesn't believe in absolutely unregulated capitalism) that makes the Two Minute Hate look like a love-in. I'm not going to pan it completely, it's a decent branch of existentialism, but it's not well-known for being very self-critical or indeed taking criticism of any sort gracefully. More of a culture problem than a belief one.
Again, her distinctly populist philosophy is called Objectivism, and thus you get names like The Objectivist Center. TOC (nee IOS, or Institute for Objectivism Studies) is actually one of the more moderate groups, because Rand's "official" legacy is carried on by a pinhead by the name of Leonard Piekoff of the Ayn Rand Institute who, to put it charitably, is nuttier than a fucking crate of baklava. Strident and dogmatic doesn't begin to describe Piekoff... but those internal politics are another story.
Anyway, obLinks:
Ayn Rand Institute
The Objectivist Center
opposing viewpoints aren't terribly well-organized, but my favorite is a paper called The Unlikeliest Cult In History. I suggest reading it only after reading one of the books (I'm sure you can do a very uncapitalist thing and download a copy over p2p), since you won't really know where it's coming from otherwise.
A broader look at the subject would show a parallel with a more modern topic: anti-depression medications. There are plenty of examples of highly creative people -- geniuses in their fields -- whose creativity would likely have been quashed if they'd had access to a good Selective Seratonin Reuptake Inhibitor. Poet Emily Dickinson [washingtonpost.com] and artist Vincent Van Gogh [vangoghgallery.com] come to mind, but I'm sure there are many others.
I'm so glad Virginia Woolf and Van Gogh were suicidal, otherwise they just wouldn't have been any use to me in creating all those great works. They died at just the right time too, none of that mediocrity that might have resulted from happiness should they have found it.
For every one of your van goghs, there's thousands of geniuses with creative potentials who spend their days in a miasma of lethargy and isolation, creating nothing, because of depression. Maybe it's just the bipolar folks who are the most entertaining...
I'm no fan of the overmedication going on, but I think you better start taking a critical eye toward your own judgements.
> Laser beams and ion canons only to get pulverized because they're not defended against a good ol' fashioned shelling.
Methinks a good megawatt laser would do a pretty decent job blowing up said incoming shells, no? My guess is they use the beam weapon to replace those old phalanx point defense guns, which do a decent job against missiles, but really have a pitifully short range.
The RPM/DEb ideas are really good. The main problem i have however , and don't really know how can be solved is combining binary packages with source code packages (eg. when you compile you own X).
.debs, but it's pretty cumbersome to do.
The FreeBSD ports system solved this a long time ago. Its dependency manager is based on make. It has a library of commands in sub-makefiles to check for things like libraries being installed, so for example when you install an app that requires qt, it checks for libqt.so using commands like ldconfig. It doesn't check for the package, it simply checks for the actual files that it depends on, and if they're not present, it builds and installs the port that contains the dependency. And ports are installed as packages and can be managed with the package manager. You get the best of both worlds. What's great about ports is that if you want to tweak something in the makefile, like options, or in the source, you just go into the source subdir, do it, then go back to the port and "make reinstall". You get your package built just the way you want.
RPM is capable of file-based dependencies instead of package-based, but the way RPM's have been managed, they typically don't (or they get it woefully wrong, as I've seen with RPM's looking for perl in a specific and wrong place, and not even trying a "which perl"). Thus you get the web of dependencies problem with RPM. debs are similar, but dpkg and apt manage them so well, there's rarely a real problem. There's a facility to build "fake"
When the time comes and you want to update , let say Libc, then you will be unable to do so because the dependecies include almost every package.
Library versioning is typically resolved by simply keeping the old version around and installing the new package side-by-side with the old one.