The MPL used by Mozilla is "more free" in the sense that it places less restrictions on the user of the code than does the GPL. It is quite fine, as far as I know, to use GPL code with the MPL and vice versa - though the work, distributed as a whole, would then need to comply with the terms of the GPL.
Nothing stops me writing a GPL plug-in for Mozilla, and so long as Mozilla don't ship it as part of one of the apps it doesn't affect Mozilla's licensing at all. This could be a perfectly sensible approach for Exchange support - though not being able to bundle support would certainly be troublesome.
For all I know, perhaps it *could* be bundled. I'm no lawyer, license or otherwise.
It's a very handy feature - plus-addressing is supported by several UNIX MTAs and MDAs, and has been for some time. I've been using it with Cyrus IMAPd at work rather heavily. Google has just adopted a little known but common mail feature, which is nice.
The utility of plus-addressing is limited by the fact that many list managers and other tools don't understand that username+mailbox@blah is the same user as user@blah , and most MUAs don't let you add an abitrary plus-addressed part to your From: address. Even so, plus-addressing combined with Sieve filtering is very handy for temporary addresses.
IPv4 address scarcity isn't artificial; it's a real issue. IPv6 will resolve it... if anybody can ever be convinced to deploy it. I support IPv6 on my home network and the network at work, and the only IPv6 communication I do is between the two sites. Nothing else I ever encounter supports it, either via 6to4 or native IPv6.
Domain name scarcity isn't artificial either. There are an essentially infinite number of domains, but they vary in desirability, and each individual domain name is unique. Domain names are not a uniform good, and are not perfectly subsitutable. I don't know about you, but my employer (the POST Newspapers, Western Australia) wouldn't particularly want http://aik0mahhievaijeisiephootequahveiraigiemeebi xietiemeeriehohhahngukashaisaathiedoorelahsukujehe shu.com.au/, though it is in fact a legal and valid domain name.
I agree that there are certainly problems with how domain names are managed, especially for the.com,.net, and.org TLDs. In these TLDs squatting and speculation is a real problem, but that's not because of artificial value. The value of a domain is very real. That doesn't, however, mean that domains must be treated as property, and it's from there that many of the issues stem.
Consider the.au TLD space. In the.au space, if you register a.com name, you should do so with a legitimate claim to the name according to the AuDA rules. If you register a name to squat (speculating, asking money from the company whose registered business name matches the domain, using the name to badmouth them, etc), the domain will simply be removed from you and handed over. You can't buy the domain, only usage rights to it. Similarly, the.org space is limited to registered charities, non-profit organizations, and so on. The.au space is well managed and suffers few or none of the problems afflicing.com,.net, and.org . This is in no small part because in.au domains are not property that you can own.
I see little hope of change in the.com,.net, and.org space. I'm personally hoping they'll just quietly die off in favour of ccTLD subdomains, but if anything the opposite appears to be happening. Too many people only understand ".com". This makes the scarcity problem worse, and does so in the very space with the worst problems. Still, what's to be done - resume the domains already "sold"?
This rule needs to go double for geeks. If the geek in question, or organization, has any history of committing the crime known as "the creation of a recursive acronym after 1985," the penalities for violation must be swift and extreme.
I admin an outfit with some 40 employees single handed. This includes all server admin, desktop support, most software purchasing and upgrades, and a fair bit of in-house development into the bargain. Full time, I expect I'd be able to handle twice that many users... so 100 isn't that much of a stretch.
That said, I suspect a desktop support tech would save them money at that load level by letting their admin focus on more important things.
At least with a single-player game, you can decide if you want to apply the next patch/update/enhancement or not.... unless you play Half Life 2.
Good game, but an apalling resource hog with an unpleasant habit of spending 10 mins updating its self, like it or not.
I hear people claim that MS bundle up multiple fixes and updates in patches, and I'm yet to see evidence of it. In fairness, I haven't really gone looking, but it also doesn't seem logical.
If MS was to bundle other (security) fixes in a patch, they would quickly be identified by reverse engineering the patch and used to exploit as-yet-unpatched systems. There are people who look over these patches in extreme detail, both "white hat" and "black hat" types.
If they bundled other fixes / changes, their business customers would get really, really pissed in a major hurry. Microsoft does NOT want to piss these people off, even with the lock they have on the market. Remember that Microsoft's whole sales pitch right now is about "total cost of ownership."
Given this, I'm inclined to belive the "MS bundles other crap with patches" rumour to be most likely outdated. It could also be something that grew out of a misunderstanding of the difference between security patches, hotfixes, and service packs. I'm more inclined to attribute breakage to a combination of (a) imperfect patch QA and (b) badly written software / malware replacing or patching system DLLs/installing drivers that end up being incompatible with "clean" versions of some of those DLLs installed by a patch. Breakage also used to be common causes of breakage in win9x... which was a horrific mess you could break by looking at it funny.
I've personally never had issues patching an NT-derived system. I ensure they're clean before patching, and I don't use shoddy software ( in so far as is possible ). In fairness, my only Windows server is NT4 (ugh); I'm speaking mostly about the XP desktops I admin at work and the older win2k machines I've run.
That's not to say that things don't go wrong for anybody, of couse... just that in my own experience they don't tend to do so. Perhaps I'm just lucky not to use $BLAH_POPULAR_DATABASE that likes to patch ntfs.sys, or whatever other ghastly hack people might perpetrate.
That works, but also disables other window manager features that it's not necessary to disable. It also pisses off the users, who find wireframe window moves and the other changes confusing and annoying.
Given that many patches have been sent in to solve this issue, it's beyond me why there's such resistance to permitting a system admin the level of control they need.
Moving windows is fine, opaque mode or wireframe. Aren't modern video cards great?
The issue is the window maximization/minimzation animation, which is the wireframe I was referring to in case that was unclrear. That's the one that causes the problems, and the one I don't understand why it can't be turned off. Everything else the window manager does is fine over remote X, it's just that damn animation. It's a small thing, for sure, but it's a case where people have repeatedly submitted small patches that make it configurable, because lots of people need this, and they're all being ignored.
If I turn on limited resources mode (which it took a lot of arguing to get included), it turns off useful things as well, including things like opaque move/resize. I then get complaints from users. It seems silly to prevent the admin from just turning off the problem feature and being done with the issue.
This particular issue drives me foaming nuts. Why? BECAUSE I USE THIN CLIENTS. That godamm wireframe is the most annying misfeature in any window manager when used remotely - it's REALLY slow, and REALLY annoying. Needing to download half of GNOME's source to fix the problem, then needing to re-do it whenever there's a security update, is way more so.
Flying can be harder, but it's mostly somewhat less time-critical, and there are a lot fewer people trying to kill you. It's unlikely that an idiot will walk out in front of your plane, and the traffic lights aren't likely to change at a bad time. Flying requires constant small corrections, but otherwise much less active interaction just to keep you alive. Above all else, you're much less likely to kill someone else if you screw up.
That said, it's no absolute either while flying or driving. It does depend on the situation, and driver skill. I'm in favour of those who argue that driving is already insanely dangerous enough, and anything that makes it worse - especially if it increases the risk of a driver killing others - just isn't OK. Don't tell me you haven't noticed the rather high proportion of morons doing stupid things on the road who're using phones vs those who are not, given the rather lower level of phone use across all drivers.
I think his concern is most likely with units that have a built-in PSU. I'd want to be careful with these too. PSUs and CRT monitors are the two things I won't work on for safety reasons.
It's not just women who're annoyed and disappointed by many games' portrayal of female characters. I find it a piss-off myself as a guy, though probably not as much.
For one thing, it's not very interesting. "Hi, I'll be your new sidekick! I'm blond, absurdly proportioned, have a stupid voice and evidently have as much brains as a pea!" Woo. I'm overblowing things, of course, but sometimes that's just how it seems, and mostly it's boring.
Second, it's often somewhat insulting. Do game designers really think that blocky images of badly animated, terribly voice acted absurd looking women with very little on (no matter how absurd and out of place) might push their game sales up much? I'd personally love to believe they'd drive them DOWN, but I fear that may be hoping too much. Leaving aside the quality of their work, there's the simple issue that it's somewhat insulting to have the game developer assuming that you're so... dumb. And hormone-driven. And tasteless. It's like we're all treated like particularly horny introverted teenagers. Yuk.
Occasionally I see an interesting character in a game. Even more occasionally a female character who isn't obviously in there as mobile scenery or to be "rescued." It's always a pleasant surprise.
I was particularly impressed with Half Life 2 in this regard. While the main female character, Alex, is... well, a bit overdone, the game mostly doesn't suffer from the issues I listed above. The peripheral characters are all particularly cool, and gender is really a side issue. Wow, a minor female character with a few interesting lines, civilized clothing that fits the scene, and a reason to be there. Heck, with "Alex" my first thought was "heh, it must be cold, that's one damn nice big jacket".
Alas, most games still suffer from having female characters appear "because they're female" not because they're characters with some reason to be there. Many also suffer from the "chain mail bikini effect" also so often observed in film. It's evident that armour worn by a woman magically becomes more effective the less of it there is, for how else could we explain such incredibly absurd and skimpy "armour" as so often appears? In games, each improved armour variant usually has even less actual armour.
Gah, I'll quit my ranting. Suffice it to say that it's more than women who're pissed off by this stuff.
To be strictly accurate, I'd say that being against porn isn't to be against freedom. However, to be in favour of imposing your own views on everybody else through censorship and regulation is, in fact, to reduce others' freedom.
In truth, we all accept that freedom cannot be complete. That would be anarchy. I would be quite free to simply kill some fellow who irritated me (though I would not) and his friends/family would be quite free to return the favour. We all accept some restrictions on our freedom in exchange for a functional society in which we can go about our lives - the argument is about where the line where restrictions cease to be appropriate is. Much as we don't have total freedom of action, we've also relinquished total freedom of speech in the form of defamation and slander laws among other things. There is no black and white issue here, that's why everybody's arguing about where the line in the grey should be drawn.
For that reason, it's not inherently a contradiction to both support "bringing freedom" to Iraq, and taking it away from USA citizens, or vice versa. That doesn't make it correct or right, of course, on either side of the matter.
That said, I'm personally of the view that the US society appears to be on a move toward dramatic restrictions in personal freedoms of speech and action, combined with reduced guarantees of safety from Governmnent abuse of power. The same is happening in Australia, where I live, a county where the majority of the population favours aping the US, something that we do blindly and incompetently but continuously. Similarly, both countries are involved in Iraq, in a war that I can see the arguments for, but am unable to morally justify (what, we can just invade whoever we feel like because we don't like them?), and am apalled by the deceptions used in the initiation of.
As for porn, I just don't understand why anybody cares. Kids who want it have always got hold of it. I'm personally really disturbed by some porn, but overall I don't really see the harm, and would frankly prefer that a kid who's going to find it anyway be able to find some that's not really gross (violent, unwilling, etc). Of course, the people wanting to stamp out porn are the same people behind abstainance-only education, and look how well that's working. *sigh*.
While that's true, it's also often worth preferring the vendor's offerings where possible. For example, if I found that I had an admin who was trying to patch MySQL by pulling down the latest upstream version and installing that instead of the version provided by the vendor, I'd be very unhappy indeed. Security patches are best handled through the vendor channel unless _extremely_ critical (think "OpenSSH remote root"), in which case I'd probably do a temporary fix until the vendor caught up.
In other cases, I'm right with you. I recently converted my core server here to Xen, splitting it into two partitions based on role. I could've upgraded to an experimental distro like FC4 to do this, but it was quicker and safer (thus cheaper) to retrofit Xen into Debian 3.1 .
Linux gives you the freedom to use vendor supplied options where it best fits your needs, and to do it yourself where that's better for you. Knowing which to pick isn't always easy, but if you make the right decisions it goes very well indeed.
... This is an incredibly contrived situation; one can simply stop and re-start the process in question after the upgrade has completed.
That is not true for a kernel patch. It's sometimes possible to unload and load a module, but most often the system must be rebooted. With vendor kernel security upgrades that's always the case, as you can't assume they've only fixed that security hole - they might've done something else, no matter how trivial, that could break compatibility of modules with your kernel.
For that matter, why didn't they just use Xen to virtualize the system into partitions? Free, easy, works with most distros released in the last few years.
That's fair. To an extent I was mixing my development oriented concerns with my issues managing Mac OS X as a system administrator.
I have the misfortune to have to administrate a SCO OpenServer box, so I'm very aware that not all systems use fstab for filesystem mappings. My issue in this case is that Mac OS X _does_ have an fstab, is based on an OS that uses fstab, documents the behaviour of fstab, and actually behaves entirely differently to how the admin might expect and what the documentation says. This sort of issue is rife throughout the OS.
As for testing, I test on what I can regularly use. That's currently two Linux distros (Debian, FC4), win32, and Mac OS X when I can endure the build times on the slow eMac I have to use remotely to do it. Due to the advantages of working in a team on open source code, the code is regularly tested on Mac OS X by a few others, plus Solaris 10/SPARC, Solaris 10/x86, pretty much any Linux distro you can name, and FreeBSD (I haven't had any reports from the other BSDs). It's not regularly tested on, but has worked fine in the past on, IRIX, HP-UX (man THAT was awful), and a large AIX machine. I'm going to be doing test builds on FreeBSD regularly soon, as I have a Xen image just about ready. Sun is working on a Xen port for Solaris 10 that'll let me do the same on Solaris x86. With Apple's move to x86 I think it'd be interesting to be able to do the same with Mac OS X.
When it comes to portability, you're quite right in that most UNIXes have many "fun" surprises. I only care about limited portability with some significant restrictions: I don't care about supporting each UNIX vendor's individual compiler (just gcc, MSVC++, and other major compilers where practical), I don't care about prehistoric UNIXes, and I'm willing to require people to use GNU make. Since most of the time I'm working on a fairly large GUI C++ app based on Qt, the goal is really portability to the/major/ UNIXes, plus Mac OS X and win32. As such, you're quite right, I'm not writing truly portable code, nor am I interested in doing so. I'm lucky enough not to have to worry about sysadmin tasks like managing network mounts with my coding, otherwise I'd consider myself to be in a very special kind of hell.
My original post mashed these two together rather more than I should have, giving the impression that my interest in testing code on Mac OS X involved these sorts of sysadmin tasks. That's thankfully not the case.
Some games that REALLY need a remake with polish, expansion, and modernized graphics but much the same core gameplay ideas:
- UFO: Enemy Unknown / X-Com: UFO Defense / X-Com: Terror from the Deep (utterly, totally, completely awesome games begging for a version that doesn't need DOS, 320x200 graphics, and a few annoying bugs) - Master of Orion II (MOO3 was barely even a game) - System Shock II (already has updated graphics and co-op mod, but co-op is a tad flakey and it barely runs on modern OSes)
Maybe Star Control II as well, though it's been updated to run on modern systems and is free now ( http://sc2.sf.net/ ). Great single player campaign.
I'd say the Monkey Island games too, but really they just want an engine port to an OS from this century. I can't imagine how you could even fix up the graphics without ruining the game.
Then, of course, there are some that've got updated versions that don't suck (eg the Civ games).
So, let me echo the sentiments of the others here - "what do you mean, old school?". Hell, the ones I've listed are relatively modern too. I'll be there are a few folks out there begging for more Commander Keen games, and then there's the MAME crowd...
First, I should note that I don't use Mac OS X much. Every time I try, it hurts - because I'm a long-time UNIX (mostly Linux) user, and nothing works even remotely "right" to my eyes. Consequently, I could easily be seeing things from the same perspective as (say) a win32 user just exposed to Linux - "oh god it's all scary and different and weird and argh!".
So, about your questions:
(1) If anything, they're moving further away from being UNIX-like with every release. Aqua, NetInfo, Launchd, etc etc etc. It's turning more and more into a whole new OS that has bits that look and work superficially like UNIX, instead of being a whole new OS with a UNIX-like subsystem that it uses for parts of its functionality. That's not a bad thing, so long as nobody tries to call Mac OS X "UNIX", though I do wish they'd do a better job of documenting what they do (or integrating it into the existing docs - a "see also" here and there in the man pages would go a long way) and making it all fit together neatly. So far, Mac OS X looks like it's an OS 1/2 way through being totally re-written, and each release includes more of the "new" OS and backwards compatibility kludges for the old one.
(2) Almost all of what they do appears to be their effort to "make things better". Sometimes I suspect they're fixing one problem with a half-assed bodge that'll cause other problems, or fixing it in a way that could arguably be done much less distruptively. None of their fixes feel very UNIX-like. (a) they're not always horrific hacks, (b) they all need special domain-specific tools, (c) the tools seem to SUCK, and (d) the changes outpace the documentation. OK, those last two are very UNIX-like;-) . At other times, they're fixing a problem really well in a way that the UNIXes can't due to the constraints of history, inertia, and backward compatibility. Some of the time, they're doing both, and almost all the time there's a hefty dose of not-invented-here syndrome going on. Also, it's clear that Apple don't believe anybody reads man pages.
I guess they're trying to make Mac OS X into a more user friendly OS, but they seem to be paying little attention to the administrator as they go.
As an example of how they don't seem to be trying to be all that sysadmin friendly, check out the section on xinetd in the launchd summary: http://developer.apple.com/macosx/launchd.html . Nice document, nice idea re launchd, but if I ever have to configure that by hand instead of xinetd I think I'll cry . It's evidently made to be configured using launchd specific utilities, which is a trend I intensely dislike but can live with - until the tools break, as tools always do.
The single biggest things that I like about UNIX (mainly BSD / Linux) are the things Apple are taking away:
- The ability to see (relatively) clearly what the system does as it performs a task / the ability to follow what the system is doing as it does it. Solaris 10 is AMAZING in this regard with dtrace, and most UNIXes are pretty good thanks to the fact that much of the OS is human readable scripts, and thanks to tools like strace/truss and gdb. - The ability to easily configure the system and alter the system's behaviour using nothing but a text editor, including not only config files but system scripts .
It's that last one that Mac OS X suffers for the loss of, and the one I most miss when I have to use it.
I should note that almost all of this is really from the sysadmin's perspective. Things are quite a bit more stable when it comes to programming APIs and ABIs (more so than your average Linux distro), and quite a bit better documented too. There are certainly plenty of quirks, but they're not/changing/ quite as fast.
The main app I work on, Scribus, has quite a lot of people who want to use it on Mac OS X. While I don't want to use it on Mac OS X , or use Mac OS X myself, I'd like to be able to make that easier for those people. Currently only one person on the team can do much work on the Mac OS X port due to issues with access to the OS and hardware. I'd like to, at the very least, be able to test against Mac OS X when a bug report comes in, and autobuild on it to make sure that no recent changes broke the Mac OS X port.
I'll work on Mac OS X for the same reason I work on autohell - it's unpleasant, but I learn a lot, help some folks out, and it helps me iron out kinks in areas that I do care more about.
I'm pretty sure I'm not the only one in this position. A lot of other OSS developers are likely to be to some extent or another. I'd be surprised if many smaller commercial developers wouldn't mind the ability to use their existing hardware to build on Mac OS X without worrying about regular software updates and introducing another operating system into the network admin's life too.
Portable coding is becoming more common (well, outside the win32 world). Portable GUI toolkits are getting more popular and they're getting better. Surely it'd be in Apple's interests to lower the barrier of entry for developers who might target Mac OS X as a secondary platform, or just want to give it a go with their app before deciding?
I also wouldn't mind being able to deploy Scribus at work at some point, and that's very unlikely to be happening on Linux for our main production users. Easy access to Mac OS X for development would make this more practical, and make me more inclined to select Mac OS X over Windows when we ditch our current Mac OS 9 machines (with pleasure, I might add).
Now, if a "developer version" would be some enormous project, then my suggestion would be very silly. Everything I've seen about Mac OS X suggests that it should be a week or two's work for a couple of people to get most of the work required done, since Mac OS X appears to be pretty modular. Most of what would be required would be removing applications, etc, which is one of the easiest things to do on a Mac.
Any quality large UPS (generally >10KVa) will be a true on-line double-conversion unit. Even in normal operation, these trickle-charge the battery, and run their load off the battery. There is zero cut-over time on AC fail and on AC restore, and running off batteries provides significant power conditioning.
These systems DO convert AC->DC, then DC->AC, 24/7.
Slashdot Mirror
The MPL used by Mozilla is "more free" in the sense that it places less restrictions on the user of the code than does the GPL. It is quite fine, as far as I know, to use GPL code with the MPL and vice versa - though the work, distributed as a whole, would then need to comply with the terms of the GPL.
Nothing stops me writing a GPL plug-in for Mozilla, and so long as Mozilla don't ship it as part of one of the apps it doesn't affect Mozilla's licensing at all. This could be a perfectly sensible approach for Exchange support - though not being able to bundle support would certainly be troublesome.
For all I know, perhaps it *could* be bundled. I'm no lawyer, license or otherwise.
It's a very handy feature - plus-addressing is supported by several UNIX MTAs and MDAs, and has been for some time. I've been using it with Cyrus IMAPd at work rather heavily. Google has just adopted a little known but common mail feature, which is nice.
The utility of plus-addressing is limited by the fact that many list managers and other tools don't understand that username+mailbox@blah is the same user as user@blah , and most MUAs don't let you add an abitrary plus-addressed part to your From: address. Even so, plus-addressing combined with Sieve filtering is very handy for temporary addresses.
BSD Jails, and Solairis zones, are two well-tested options in this area.
IPv4 address scarcity isn't artificial; it's a real issue. IPv6 will resolve it ... if anybody can ever be convinced to deploy it. I support IPv6 on my home network and the network at work, and the only IPv6 communication I do is between the two sites. Nothing else I ever encounter supports it, either via 6to4 or native IPv6.
i xietiemeeriehohhahngukashaisaathiedoorelahsukujehe shu.com.au/, though it is in fact a legal and valid domain name.
.com, .net, and .org TLDs. In these TLDs squatting and speculation is a real problem, but that's not because of artificial value. The value of a domain is very real. That doesn't, however, mean that domains must be treated as property, and it's from there that many of the issues stem.
.au TLD space. In the .au space, if you register a .com name, you should do so with a legitimate claim to the name according to the AuDA rules. If you register a name to squat (speculating, asking money from the company whose registered business name matches the domain, using the name to badmouth them, etc), the domain will simply be removed from you and handed over. You can't buy the domain, only usage rights to it. Similarly, the .org space is limited to registered charities, non-profit organizations, and so on. The .au space is well managed and suffers few or none of the problems afflicing .com, .net, and .org . This is in no small part because in .au domains are not property that you can own.
.com, .net, and .org space. I'm personally hoping they'll just quietly die off in favour of ccTLD subdomains, but if anything the opposite appears to be happening. Too many people only understand ".com". This makes the scarcity problem worse, and does so in the very space with the worst problems. Still, what's to be done - resume the domains already "sold"?
Domain name scarcity isn't artificial either. There are an essentially infinite number of domains, but they vary in desirability, and each individual domain name is unique. Domain names are not a uniform good, and are not perfectly subsitutable. I don't know about you, but my employer (the POST Newspapers, Western Australia) wouldn't particularly want http://aik0mahhievaijeisiephootequahveiraigiemeeb
I agree that there are certainly problems with how domain names are managed, especially for the
Consider the
I see little hope of change in the
This rule needs to go double for geeks. If the geek in question, or organization, has any history of committing the crime known as "the creation of a recursive acronym after 1985," the penalities for violation must be swift and extreme.
I'm not sure 1:100 is so bad.
... so 100 isn't that much of a stretch.
I admin an outfit with some 40 employees single handed. This includes all server admin, desktop support, most software purchasing and upgrades, and a fair bit of in-house development into the bargain. Full time, I expect I'd be able to handle twice that many users
That said, I suspect a desktop support tech would save them money at that load level by letting their admin focus on more important things.
At least with a single-player game, you can decide if you want to apply the next patch/update/enhancement or not. ... unless you play Half Life 2.
Good game, but an apalling resource hog with an unpleasant habit of spending 10 mins updating its self, like it or not.
I hear people claim that MS bundle up multiple fixes and updates in patches, and I'm yet to see evidence of it. In fairness, I haven't really gone looking, but it also doesn't seem logical.
... which was a horrific mess you could break by looking at it funny.
If MS was to bundle other (security) fixes in a patch, they would quickly be identified by reverse engineering the patch and used to exploit as-yet-unpatched systems. There are people who look over these patches in extreme detail, both "white hat" and "black hat" types.
If they bundled other fixes / changes, their business customers would get really, really pissed in a major hurry. Microsoft does NOT want to piss these people off, even with the lock they have on the market. Remember that Microsoft's whole sales pitch right now is about "total cost of ownership."
Given this, I'm inclined to belive the "MS bundles other crap with patches" rumour to be most likely outdated. It could also be something that grew out of a misunderstanding of the difference between security patches, hotfixes, and service packs. I'm more inclined to attribute breakage to a combination of (a) imperfect patch QA and (b) badly written software / malware replacing or patching system DLLs/installing drivers that end up being incompatible with "clean" versions of some of those DLLs installed by a patch. Breakage also used to be common causes of breakage in win9x
I've personally never had issues patching an NT-derived system. I ensure they're clean before patching, and I don't use shoddy software ( in so far as is possible ). In fairness, my only Windows server is NT4 (ugh); I'm speaking mostly about the XP desktops I admin at work and the older win2k machines I've run.
That's not to say that things don't go wrong for anybody, of couse... just that in my own experience they don't tend to do so. Perhaps I'm just lucky not to use $BLAH_POPULAR_DATABASE that likes to patch ntfs.sys, or whatever other ghastly hack people might perpetrate.
"This legislation proudly bought to you by EMC, NetApp, and Sun"
Probably not, but right now I do suspect those three will be partying hard.
That works, but also disables other window manager features that it's not necessary to disable. It also pisses off the users, who find wireframe window moves and the other changes confusing and annoying.
Given that many patches have been sent in to solve this issue, it's beyond me why there's such resistance to permitting a system admin the level of control they need.
Moving windows is fine, opaque mode or wireframe. Aren't modern video cards great?
The issue is the window maximization/minimzation animation, which is the wireframe I was referring to in case that was unclrear. That's the one that causes the problems, and the one I don't understand why it can't be turned off. Everything else the window manager does is fine over remote X, it's just that damn animation. It's a small thing, for sure, but it's a case where people have repeatedly submitted small patches that make it configurable, because lots of people need this, and they're all being ignored.
If I turn on limited resources mode (which it took a lot of arguing to get included), it turns off useful things as well, including things like opaque move/resize. I then get complaints from users. It seems silly to prevent the admin from just turning off the problem feature and being done with the issue.
This particular issue drives me foaming nuts. Why? BECAUSE I USE THIN CLIENTS. That godamm wireframe is the most annying misfeature in any window manager when used remotely - it's REALLY slow, and REALLY annoying. Needing to download half of GNOME's source to fix the problem, then needing to re-do it whenever there's a security update, is way more so.
What the heck is wrong with a gconf key? ARGH.
</rant>
Flying can be harder, but it's mostly somewhat less time-critical, and there are a lot fewer people trying to kill you. It's unlikely that an idiot will walk out in front of your plane, and the traffic lights aren't likely to change at a bad time. Flying requires constant small corrections, but otherwise much less active interaction just to keep you alive. Above all else, you're much less likely to kill someone else if you screw up.
That said, it's no absolute either while flying or driving. It does depend on the situation, and driver skill. I'm in favour of those who argue that driving is already insanely dangerous enough, and anything that makes it worse - especially if it increases the risk of a driver killing others - just isn't OK. Don't tell me you haven't noticed the rather high proportion of morons doing stupid things on the road who're using phones vs those who are not, given the rather lower level of phone use across all drivers.
I think his concern is most likely with units that have a built-in PSU. I'd want to be careful with these too. PSUs and CRT monitors are the two things I won't work on for safety reasons.
It's not just women who're annoyed and disappointed by many games' portrayal of female characters. I find it a piss-off myself as a guy, though probably not as much.
... dumb. And hormone-driven. And tasteless. It's like we're all treated like particularly horny introverted teenagers. Yuk.
... well, a bit overdone, the game mostly doesn't suffer from the issues I listed above. The peripheral characters are all particularly cool, and gender is really a side issue. Wow, a minor female character with a few interesting lines, civilized clothing that fits the scene, and a reason to be there. Heck, with "Alex" my first thought was "heh, it must be cold, that's one damn nice big jacket".
For one thing, it's not very interesting. "Hi, I'll be your new sidekick! I'm blond, absurdly proportioned, have a stupid voice and evidently have as much brains as a pea!" Woo. I'm overblowing things, of course, but sometimes that's just how it seems, and mostly it's boring.
Second, it's often somewhat insulting. Do game designers really think that blocky images of badly animated, terribly voice acted absurd looking women with very little on (no matter how absurd and out of place) might push their game sales up much? I'd personally love to believe they'd drive them DOWN, but I fear that may be hoping too much. Leaving aside the quality of their work, there's the simple issue that it's somewhat insulting to have the game developer assuming that you're so
Occasionally I see an interesting character in a game. Even more occasionally a female character who isn't obviously in there as mobile scenery or to be "rescued." It's always a pleasant surprise.
I was particularly impressed with Half Life 2 in this regard. While the main female character, Alex, is
Alas, most games still suffer from having female characters appear "because they're female" not because they're characters with some reason to be there. Many also suffer from the "chain mail bikini effect" also so often observed in film. It's evident that armour worn by a woman magically becomes more effective the less of it there is, for how else could we explain such incredibly absurd and skimpy "armour" as so often appears? In games, each improved armour variant usually has even less actual armour.
Gah, I'll quit my ranting. Suffice it to say that it's more than women who're pissed off by this stuff.
To be strictly accurate, I'd say that being against porn isn't to be against freedom. However, to be in favour of imposing your own views on everybody else through censorship and regulation is, in fact, to reduce others' freedom.
In truth, we all accept that freedom cannot be complete. That would be anarchy. I would be quite free to simply kill some fellow who irritated me (though I would not) and his friends/family would be quite free to return the favour. We all accept some restrictions on our freedom in exchange for a functional society in which we can go about our lives - the argument is about where the line where restrictions cease to be appropriate is. Much as we don't have total freedom of action, we've also relinquished total freedom of speech in the form of defamation and slander laws among other things. There is no black and white issue here, that's why everybody's arguing about where the line in the grey should be drawn.
For that reason, it's not inherently a contradiction to both support "bringing freedom" to Iraq, and taking it away from USA citizens, or vice versa. That doesn't make it correct or right, of course, on either side of the matter.
That said, I'm personally of the view that the US society appears to be on a move toward dramatic restrictions in personal freedoms of speech and action, combined with reduced guarantees of safety from Governmnent abuse of power. The same is happening in Australia, where I live, a county where the majority of the population favours aping the US, something that we do blindly and incompetently but continuously. Similarly, both countries are involved in Iraq, in a war that I can see the arguments for, but am unable to morally justify (what, we can just invade whoever we feel like because we don't like them?), and am apalled by the deceptions used in the initiation of.
As for porn, I just don't understand why anybody cares. Kids who want it have always got hold of it. I'm personally really disturbed by some porn, but overall I don't really see the harm, and would frankly prefer that a kid who's going to find it anyway be able to find some that's not really gross (violent, unwilling, etc). Of course, the people wanting to stamp out porn are the same people behind abstainance-only education, and look how well that's working. *sigh*.
While that's true, it's also often worth preferring the vendor's offerings where possible. For example, if I found that I had an admin who was trying to patch MySQL by pulling down the latest upstream version and installing that instead of the version provided by the vendor, I'd be very unhappy indeed. Security patches are best handled through the vendor channel unless _extremely_ critical (think "OpenSSH remote root"), in which case I'd probably do a temporary fix until the vendor caught up.
In other cases, I'm right with you. I recently converted my core server here to Xen, splitting it into two partitions based on role. I could've upgraded to an experimental distro like FC4 to do this, but it was quicker and safer (thus cheaper) to retrofit Xen into Debian 3.1 .
Linux gives you the freedom to use vendor supplied options where it best fits your needs, and to do it yourself where that's better for you. Knowing which to pick isn't always easy, but if you make the right decisions it goes very well indeed.
That is not true for a kernel patch. It's sometimes possible to unload and load a module, but most often the system must be rebooted. With vendor kernel security upgrades that's always the case, as you can't assume they've only fixed that security hole - they might've done something else, no matter how trivial, that could break compatibility of modules with your kernel.
In most cases, a restart is the only sane option.
For that matter, why didn't they just use Xen to virtualize the system into partitions? Free, easy, works with most distros released in the last few years.
Well, I'm impressed. First post, and it wasn't "first post!".
That's fair. To an extent I was mixing my development oriented concerns with my issues managing Mac OS X as a system administrator.
/major/ UNIXes, plus Mac OS X and win32. As such, you're quite right, I'm not writing truly portable code, nor am I interested in doing so. I'm lucky enough not to have to worry about sysadmin tasks like managing network mounts with my coding, otherwise I'd consider myself to be in a very special kind of hell.
I have the misfortune to have to administrate a SCO OpenServer box, so I'm very aware that not all systems use fstab for filesystem mappings. My issue in this case is that Mac OS X _does_ have an fstab, is based on an OS that uses fstab, documents the behaviour of fstab, and actually behaves entirely differently to how the admin might expect and what the documentation says. This sort of issue is rife throughout the OS.
As for testing, I test on what I can regularly use. That's currently two Linux distros (Debian, FC4), win32, and Mac OS X when I can endure the build times on the slow eMac I have to use remotely to do it. Due to the advantages of working in a team on open source code, the code is regularly tested on Mac OS X by a few others, plus Solaris 10/SPARC, Solaris 10/x86, pretty much any Linux distro you can name, and FreeBSD (I haven't had any reports from the other BSDs). It's not regularly tested on, but has worked fine in the past on, IRIX, HP-UX (man THAT was awful), and a large AIX machine. I'm going to be doing test builds on FreeBSD regularly soon, as I have a Xen image just about ready. Sun is working on a Xen port for Solaris 10 that'll let me do the same on Solaris x86. With Apple's move to x86 I think it'd be interesting to be able to do the same with Mac OS X.
When it comes to portability, you're quite right in that most UNIXes have many "fun" surprises. I only care about limited portability with some significant restrictions: I don't care about supporting each UNIX vendor's individual compiler (just gcc, MSVC++, and other major compilers where practical), I don't care about prehistoric UNIXes, and I'm willing to require people to use GNU make. Since most of the time I'm working on a fairly large GUI C++ app based on Qt, the goal is really portability to the
My original post mashed these two together rather more than I should have, giving the impression that my interest in testing code on Mac OS X involved these sorts of sysadmin tasks. That's thankfully not the case.
Some games that REALLY need a remake with polish, expansion, and modernized graphics but much the same core gameplay ideas:
- UFO: Enemy Unknown / X-Com: UFO Defense / X-Com: Terror from the Deep (utterly, totally, completely awesome games begging for a version that doesn't need DOS, 320x200 graphics, and a few annoying bugs)
- Master of Orion II (MOO3 was barely even a game)
- System Shock II (already has updated graphics and co-op mod, but co-op is a tad flakey and it barely runs on modern OSes)
Maybe Star Control II as well, though it's been updated to run on modern systems and is free now ( http://sc2.sf.net/ ). Great single player campaign.
I'd say the Monkey Island games too, but really they just want an engine port to an OS from this century. I can't imagine how you could even fix up the graphics without ruining the game.
Then, of course, there are some that've got updated versions that don't suck (eg the Civ games).
So, let me echo the sentiments of the others here - "what do you mean, old school?". Hell, the ones I've listed are relatively modern too. I'll be there are a few folks out there begging for more Commander Keen games, and then there's the MAME crowd...
First, I should note that I don't use Mac OS X much. Every time I try, it hurts - because I'm a long-time UNIX (mostly Linux) user, and nothing works even remotely "right" to my eyes. Consequently, I could easily be seeing things from the same perspective as (say) a win32 user just exposed to Linux - "oh god it's all scary and different and weird and argh!".
;-) . At other times, they're fixing a problem really well in a way that the UNIXes can't due to the constraints of history, inertia, and backward compatibility. Some of the time, they're doing both, and almost all the time there's a hefty dose of not-invented-here syndrome going on. Also, it's clear that Apple don't believe anybody reads man pages.
/changing/ quite as fast.
So, about your questions:
(1) If anything, they're moving further away from being UNIX-like with every release. Aqua, NetInfo, Launchd, etc etc etc. It's turning more and more into a whole new OS that has bits that look and work superficially like UNIX, instead of being a whole new OS with a UNIX-like subsystem that it uses for parts of its functionality. That's not a bad thing, so long as nobody tries to call Mac OS X "UNIX", though I do wish they'd do a better job of documenting what they do (or integrating it into the existing docs - a "see also" here and there in the man pages would go a long way) and making it all fit together neatly. So far, Mac OS X looks like it's an OS 1/2 way through being totally re-written, and each release includes more of the "new" OS and backwards compatibility kludges for the old one.
(2) Almost all of what they do appears to be their effort to "make things better". Sometimes I suspect they're fixing one problem with a half-assed bodge that'll cause other problems, or fixing it in a way that could arguably be done much less distruptively. None of their fixes feel very UNIX-like. (a) they're not always horrific hacks, (b) they all need special domain-specific tools, (c) the tools seem to SUCK, and (d) the changes outpace the documentation. OK, those last two are very UNIX-like
I guess they're trying to make Mac OS X into a more user friendly OS, but they seem to be paying little attention to the administrator as they go.
As an example of how they don't seem to be trying to be all that sysadmin friendly, check out the section on xinetd in the launchd summary: http://developer.apple.com/macosx/launchd.html . Nice document, nice idea re launchd, but if I ever have to configure that by hand instead of xinetd I think I'll cry . It's evidently made to be configured using launchd specific utilities, which is a trend I intensely dislike but can live with - until the tools break, as tools always do.
The single biggest things that I like about UNIX (mainly BSD / Linux) are the things Apple are taking away:
- The ability to see (relatively) clearly what the system does as it performs a task / the ability to follow what the system is doing as it does it. Solaris 10 is AMAZING in this regard with dtrace, and most UNIXes are pretty good thanks to the fact that much of the OS is human readable scripts, and thanks to tools like strace/truss and gdb.
- The ability to easily configure the system and alter the system's behaviour using nothing but a text editor, including not only config files but system scripts .
It's that last one that Mac OS X suffers for the loss of, and the one I most miss when I have to use it.
I should note that almost all of this is really from the sysadmin's perspective. Things are quite a bit more stable when it comes to programming APIs and ABIs (more so than your average Linux distro), and quite a bit better documented too. There are certainly plenty of quirks, but they're not
The main app I work on, Scribus, has quite a lot of people who want to use it on Mac OS X. While I don't want to use it on Mac OS X , or use Mac OS X myself, I'd like to be able to make that easier for those people. Currently only one person on the team can do much work on the Mac OS X port due to issues with access to the OS and hardware. I'd like to, at the very least, be able to test against Mac OS X when a bug report comes in, and autobuild on it to make sure that no recent changes broke the Mac OS X port.
I'll work on Mac OS X for the same reason I work on autohell - it's unpleasant, but I learn a lot, help some folks out, and it helps me iron out kinks in areas that I do care more about.
I'm pretty sure I'm not the only one in this position. A lot of other OSS developers are likely to be to some extent or another. I'd be surprised if many smaller commercial developers wouldn't mind the ability to use their existing hardware to build on Mac OS X without worrying about regular software updates and introducing another operating system into the network admin's life too.
Portable coding is becoming more common (well, outside the win32 world). Portable GUI toolkits are getting more popular and they're getting better. Surely it'd be in Apple's interests to lower the barrier of entry for developers who might target Mac OS X as a secondary platform, or just want to give it a go with their app before deciding?
I also wouldn't mind being able to deploy Scribus at work at some point, and that's very unlikely to be happening on Linux for our main production users. Easy access to Mac OS X for development would make this more practical, and make me more inclined to select Mac OS X over Windows when we ditch our current Mac OS 9 machines (with pleasure, I might add).
Now, if a "developer version" would be some enormous project, then my suggestion would be very silly. Everything I've seen about Mac OS X suggests that it should be a week or two's work for a couple of people to get most of the work required done, since Mac OS X appears to be pretty modular. Most of what would be required would be removing applications, etc, which is one of the easiest things to do on a Mac.
Any quality large UPS (generally >10KVa) will be a true on-line double-conversion unit. Even in normal operation, these trickle-charge the battery, and run their load off the battery. There is zero cut-over time on AC fail and on AC restore, and running off batteries provides significant power conditioning.
These systems DO convert AC->DC, then DC->AC, 24/7.