Funnily enough, I've been working on exactly this over the weekend. Unfortunately, the docs are terrible. Some of the Wiki info suggests that things have improved, but I'm discovering a decent-sized client cache (10Gb) so that I can offline most of what I'd use has horrendous occaisional slow-downs and pauses. I'm planning on testing a local/client server and a client with the RVM turned-off this week, but I'm not keen on the size that the RVM file(s) will have to get to. Previous comments in here suggesting it's not really designed for modern data-sets (gigs rather than megs) are starting to look as if that's true... Other than that, it actually looks like a reasonably good design!
I'd like to second all of the above. I spent a long, long time trying and testing as many OSS SCM's as I could, and have decided that Subversion is "unfortunate". I like their multiple-architecure support -- using the apache cross-platform libs was rather inspired... but that's about it!
Being able to apply tags and branches on the exact same tree is conceptually more natural, even though Subversion's zero-copy/linking initially looks better.
I'm a big fan of darcs, too, and creating new repo's as branches is easy to get used to, but a little messy when you want to compare differences/changes.
In addition to this, the Subversions multiple full-body redesign has always made me very, very nervous. I still find it hard to believe that Subversion won't undergo another major redesign at any point!
To be honest, SVK almost convinced me of using Subversion, especially as it solved Subversion's fundamental difficulties with merging, but by then, Git and darcs had proved themselves far superior.
All the A.D. advantages you've mentioned are avail. under OSS, long before A.D. had them. Perl modules for LDAP are old news. They're very useful. In fact, I've made Perl update PDC data a lot more easily than anything else I've used.
You can easily create multiple hierarchies under one LDAP Base DN, and apply your ACLs based on the Base DN. You can even create multiple Base DNs running on different servers, and teach them how to pass clients off to each one. If you really need strong auth and auth-domains, you should be looking at Kerberos. Even A.D. is based on it, and many, many products can use it (i.e. Samba, PAM libraries, etc). I'm not too sure how it interacts with LDAP, but I believe it can with little pain -- in fact I vaguely remember OpenLDAP can auth. against Kerberos somehow...
And the thing about being able to use it "out of the box" implies a workable set-up as soon as it's installed, which just doesn't happen in the real world: You read the docs/go on your course You adjust your expectations You install You configure it properly for your environs You test You add some power-users You fix You get it signed-off You deploy
(You fight off complaining users who are never satisfied)
I was involved in setting up a similar system in a prev. job. Basically, if you're expecting to use A.D anywhere, you're really advised to stick to all-MS.
We worked hard on getting A.D. to play nicely with a Unix LDAP system, Bind (DNS), Samba, etc. and it just wasn't even slightly fun. There's quite a few hacks that they use, and they seem to expect an ability to dynamically-update quite a few things (e.g. in DNS) which was tricky to get going with Unix tools. On top of that, it will be expensive.
However, if you avoid A.D, and even Windows PDC's, it's actually fairly easy. OpenLDAP is mostly only tricky for Access-Controls, Samba 3 can do pretty-much everything SMB/CIFS file/print-related, and can auth. against LDAP easily. We preferred Exim over Sendmail, Postfix, and QMail, but just pick the one you like best as they all do LDAP. We installed Dovecot for the IMAP server -- does LDAP, too.
I think the main point is: if you use some decent (read: fully-compliant) LDAP server, or X.500 + LDAP shim, the rest of it can be whatever you like best.
I would like to put in a couple of other points:
For what you're aiming for, OSS will do it all. (e.g. OpenLDAP, Samba, Exim + DSpam + ClamAV, Dovecot/Courier, SquirrelMail...). If you're prepared to give your staff time to test-drive and learn the products, it's probably money better spent rather than giving away in licenses.
Pick OSS s/w that has decent docs. I find that to be a reasonable bench-mark for both its popularity and likelihood for it to stick around.
If you don't care about OSS, I personally have had good experiences with Lotus Notes. It is fairly straight-forward to use and Admin, tries hard to use standard protocols (e.g. IMAP, LDAP, NNTP...) for non-Notes clients and the document-management abilities will make you wonder why you never thought about it before! However licenses start at £150-ish/user, and £3000-ish/server... (sorry if I mis-remembered those prices!)
Well, went there on holiday for the last fortnight, to see some friends who've been working there for almost a year.
My friends are very, very, luck, as their wages are paid by a British company, but they say most people there are unbelievably badly paid cf. Western countries, even when you take into account the difference in cost of living. e.g. a lot live in tin-shacks, just as you see in pictures on the TV of African shanty-towns, £60/month was a common wage.
Also, import-duty is extremely high, too: 95% for imported cars, for example. So you don't see a great deal of "foreign conveniences".
Part of the "scam" might be the resorts are usually stunning there. The tourist packages will never show you the run-down areas. It reminds me of the Yucatan peninsula in Mexico (though perhaps less run-down in Mauritius) where the tourist resorts are lovely and the coastal towns are not a lot worse than some English ones, but when you hop in a car or take a tourist-coach through the inland towns you realise that people are only just surviving.
However, I'm told most Mauritians are much happier and friendlier cf. Londoners even so. The speculation there was the huge quantities of alcohol drunk by the general population -- when large portions of the country are covered in sugar-cane, it's hardly surprising there's a lot of cheap (and very excellent) rum.
NB: You'll almost never see a bottle of Bicardi there because it's just not as good as the local rum, and import prices make it more expensive, anyway.
But anyway, if you're expecting to be able to have a similar life-style to a modern Western city, you'll get a rude shock. If you're wanting to get away from the "modern-ness" of the Western cities, you'll probably enjoy it.
I'd heartily agree with learning a language for slightly different reasons; though programmers logically would probably benefit the most, the advantage is in having to translate concepts into a another form in a way that is less than automatic.
I personally don't think it matters hugely which language, as they *all* have idiosms (sp?) that you have to know how to deal with.
e.g I learnt French in college and University, and it was a milestone to understand that certain phrases (some words, too, but that was more obvious) just can't translate properly, and the understanding had so much more to do with understanding the culture, traditions, history, etc than any of the words. I also learnt Maori (New Zealand) for short while, and found it bloody difficult until I stopped trying to translate any words independantly, and just worked with whole phrases.
These other effects on/of language are some of the things that I think are helpful with IC&T in general, partly because there is a strong culture and history there which dictates why some things are the way they are. e.g what does the phrase "religious war" mean to a Techie vs. non-Techie? e.g My wife would like someone to "just write a program" that'll make Hotmail work the way she wants without her having to remember the less-than-intuitive interface idiosyncracies; with her not knowing (or really wanting to know) all the detail that goes into Hotmail, the Internet, Client-Server, etc, etc, it's down to me to explain the relevant concepts in her "language" so she doesn't beat the computer into dust;-)
As an aside: it seems to me the best programmers out there are those who can pick up real-world languages with some degree of ease.
If I remember rightly, it used to be you bought the hardware (at huge expense) and the software was thrown in, including source code. The hardware was mostly useless without it, after all.
I think you were forced to buy a maintenance contract and all that, but the software wasn't considered valuable enough in it's own right for a pretty long time. I don't think many companies copyrighted it or anything, either.
And isn't that how a lot of the old-school Unix hackers (read: "current OSS advocates") got going: hacking/fixing the vendor-supplied OS?
I've not tried s/w RAID-5 under Linux, as, among other things, a Solaris DSO course I went on recommended against it so heavily -- and not just because Sun wanted to sell us more hardware;-) But anyway, I've striped partitions across Linux (LVM2) and Solaris disks (Disk Suite/Vol Manager) several times -- RAID 0 I think that is -- and that seems to be pretty efficient.
Mostly it has to do with the fact that good h/w RAID has one channel and possibly bus dedicated per disk (whether SCSI or IDE) which means there's very low access-contention, but it also helps (a lot) that the dedicated controller handles all the splitting/consolidating of the data across the disks, parity-checks when a disk is replaced, etc, etc, so the OS can get on with what its better at. Sharing a channel/bus (e.g a single IDE cable or a single PCI Bus) can really affect read/write performance because the data will have to be consolidated in a divided manner across those same channels/buses from all the disks (or all-but-one for RAID-5). The speed of those channels/buses can vary a lot between systems and architectures, though a 500 MHz CPU suggests a less-than-desirable speed for this task...
Parity-checking a replacement disk will also absolutely destroy disk-access speed -- *all* the data on *all* the disks has to be read in to write the correct parity to the new disk -- but a good h/w RAID should still allow the array to be used at the same time without too much pain.
Anyway, since you've got the disks already, it would be worthwhile playing with RAID and finding out for yourself, but if you can can't spare the time (or headaches) I'd definitely think seriously about a h/w RAID solution.
The 2 biggest reasons I gave on KDE, after trying it many times over the years, is that it still feels noticably slower, but more importantly: C++ completely locks me out of using some types of development.
C++ libraries on a C-Object design (think GObjects, etc) would probably allow me to use Qt easily and interact without some sort of middleware, while still abstracting the interface nicely.
On top of that, Qt's MOC/preprocessor style adds a further layer of cruft that I think is pretty unecessary these days.
I shouldn't knock it too hard, as it was a *good* idea at a time where the next best option was the Visual C++ on MFC style, but it's a harsh lock-in these days.
However, there does seem to be certain aspects of GNOME severely lacking. e.g. a simple Printer-widget/interface would make a big difference to me...
I personally think he has a good point about the Subversion design. VCS systems are pretty complicated beasts when it comes down to it, and deserve a fair bit of work on the design before doing any production-oriented coding. From what I remember of Subversion, looking at it on-and-off over the years, they have ended up redesigning it several times over, while trying to produce code that would end up being part of the finished product, thus throwing away large portions several times because it didn't fit the new design, rather than the code could be done in a better way. I also found it a little strange to be trying to basically implement a VC file-system with file-attributes... Why not actually make a VFS plug-in for Linux, even if it forces everyone to adopt a Linux Server for the repository, and do the merge, etc, tools at the user-level? It still seems more elegant that way, even if it forces use of of Linux -- didn't VMS have a rudimentary Rev Control in its file-system? When I first saw Arch, I immediately shied away because it was all written in shell, however when the first version of 'tla' appeared (i.e the C-code version) I was quite impressed at the fact he was actually allowing himself to test his basic desing ideas in what amounts to a Rapid Dev Environment, even if Shell-script is a painful beast to try and do something like this in.
We have Debian here. We also have Solaris, SuSE (and a little bit of RedHat and Fedora for those who don't know any better) and used to have Gentoo.
We use Debian on any Linux servers -- primarily because Yast gets in the way for most serious Unix admins -- and SuSE on most workstations -- it makes it a lot easier for Noobs who actually need to be able to 'fiddle' with their box (driver development, etc). I like them both, but prefer Debian.
Our sales pitch is based on the fact that we usually wouldn't bother calling for vendor support, anyway. As fast as some of them are, they're usually not as fast as us researching and fixing it ourselves; you have the phone-call, then sometimes the Appeasment Engineer, then they want a config. dump so that they can look at the problem, then they often come back to you wanting more information... Whereas even with a bit of Googling and some mail-list trawling, someone's often described your problem and how they fixed it. There's not a huge number of 'new' problems out there.
On top of this, we started off with small introductions -- e.g a CUPS server was much easier and faster to get up-and-running on Debian than Solaris -- generally setting up the machine as proof-of-concept, then deciding the thing worked reliably enough not to migrate it elsewhere.
I think it ultimately doesn't matter who the vendor is, they're going to be rubbish at some point or for some thing, and if you're not going to be able to present yourself as the first and fastest trouble-shooter in a large percentage of situations, almost any OS/Distribution is going to bite you hard and often.
With Linux, I would say that percentage needs to be even higher, just because the tools and skills in use by the Support Engineers are probably going to be exactly the same ones available to yourself -- i.e any of the non-open-source vendor's proprietory tools and trade-secret knowledge is in the public-domain for open-source.
However, if you're mostly wanting to make it possible to redirect blame, then there's nothing better than a vendor.
I'm a huge fan of Ruby, and have been using it for years, however I still use Perl a lot, mostly because more people know how to use it at my work, and fewer people understand objective programming here, too.
But another reason I still use Perl is 'Perl 6', and the fact it seems to be incorporating a lot of the things I like about Ruby, plus a couple of other interesting ones, but also because the Parrot VM is supposed to allow you to write different parts in Perl or Ruby (or Python, etc?) or *whatever* is appropriate at the time.
I really like the idea of your HTML developer being able to use their polished PHP/HTML skills to suck the data through the nicely objectified RubyOracle libs your DB Developer wrote (instead of the rubbish method(s) PHP insists on now for accessing SQL-ised data...) which can also interact with the more up-to-date Perl6/LDAP libs, all because Parrot lets them work together at the code-level.
I'd never heard of Maporama before.
I'm from New Zealand, but left for Britain 5 years ago. In that time, this is the first mapping web-site that even knew NZ existed, let alone could map street names -> map locations...
Obviously I've just been using the wrong web-site, as I notice the maps are from NavTech, and I believe that's one of the more popular providers to Mapping web-sites.
Sorry, I should've added 'proof of purchase' as one of the requirements, but that's mostly to avoid shop-lifters getting refunds for things they've stolen.
As far as I understood it, they have to accept returns, so long as it's within a reasonable time-limit -- I think 28 days -- and in a 'resaleable' condition -- which usually means unopened and undamaged, and sometimes completely unused, e.g under-pants, but that can sometimes be too hard to prove.
Anything above that is at the shop's discretion. I think it's also up to you to take it to court if they dispute it.
My information's mostly from TV programmes I've seen recently about kids being ripped-off because they didn't know these rights.
In Britain and New Zealand, a returns policy is mandated by law: provided the product is returned in a resaleable condition. In fact, it's probably the same in Europe and the rest of Australasia.
Several stores in the UK will allow you to return CDs with copy-protection (i.e. the variety that causes Mac CD-drives to lock up, and won't play properly in most PC CD-drives) without any questions, too, because of the problems and confusion it's caused.
I've heard New Zealanders and Aussie's are popular for certain broadcast roles in the US because when they put-on a US accent, no-one can tell where they're from, therefore there's no regional bias to the speaker.
But as to the UK news-readers having a 'neutral British' accent, that's an utter fabrication. A lot of presenters have a slightly posh South-East or Midlands accent, but it's just as likely you'll hear Northern, Irish, Scottish, Cornish, Welsh, etc, etc. It's more that no-one really cares much about the presenter's accent, only that they come across as appropriate to that TV show's audience, and that they can be understood.
As an aside: I've never heard anyone able to put-on a proper Kiwi accent who hadn't lived there for at least 2 years. It's easy to recognise an Aussie one, for example, even a Western-Australian one, if you've lived in NZ for long enough, and I'd say Aus and NZ accents are very close cousins.
I'm sure this point's been argued elsewhere, but: the whole reason for regulation and legislation over any and all employment discrimination is because the companies would otherwise always have the upper-hand in any employee-related dispute, and most companies do not act in anything like a loyal manner to their employees. This is not 'we can no longer support this position', which is a result of change and competition -- the employee also has the right to say 'I can no longer work in this position -- but more along the lines of, 'we don't care how much you've given in time, effort and expertise, while you have this condition or disability, because our accountants think you're statistically too much of a liability'. Statistics and lies, no common sense.
So many people have bought into a management style that emphasises live to work, even if it kills you, and this esoteric and unrealistic concept of 'growth,' year-on-year, forever, that there is becoming more of an insistence on employees which fit into some kind-of super-human species, or the nearest to that available, that we're ending up with stressed-out, over-tired, and genuinely worried employees instead. The idea that a young employee will be any better than an older one seems to be based on the enthusiasm of youth -- which jades pretty quickly -- and ignores its stupidity and self-centredness. The idea that a pregnant woman being 'off sick' for 6 months id too much of a liability is very short-sighted. I've heard excuses that they'll decide to stay at home for 5 years looking after kids as a housewife, anyway, but it's no longer anywhere near as accurate as 30 years ago, esp. to the modern career-woman. The idea that your culture and its stereotype/reputation will be an accurate indication of work ethics for every individual is frequently proved wrong.
And all this is supported by legislation and regulation because of the simple fact that people are too-often short-sighted, management too ruthless/over-worked/incompetent, and companies too focused on competition, profit and 'growth'. I am glad the government tries to support the people as a whole more than minorities and individuals -- incl. in the previous post's 'company' comparison to indivuals -- at the same time as trying to help minorities and individuals to survive and/or thrive. It's a balancing act that could so easily fall to bits, and despite the modern capitalist's support of letting only the strong and unemotional survive, at the expense of our humanity and neighbours, it'd be a terrible, Klingon-esque world if we didn't keep working on that balance.
I personally think Microkernels are *still* a good idea. A well-designed one is extremely modular, often extremely small (yeah, I know: "duh") and very portable, and the small-chunk approach can make them a lot easier to understand. On top of this, you can seperate the different parts of the kernel from each other, so there's far less likelihood of different pieces stepping on others, drivers can (theoretically) be run in user-space (if such a thing applies)...
However, like anything, they are *not* going to be a perfect panacea. If the design is not just right, they can end up being far worse than the mono-kernels they are often competing with. The Mach MK is one where kernel-developers complain frequently about trying to get drivers to work with it, its message-passing architecture being unable/difficult to use in certain important ways, etc, etc, whereas I've heard a lot of praise for the L4 design. In fact, isn't TRON/iTRON (etc) a MK-style design? If it weren't for the US Trade Dept, it'd probably be running on your Desktop, PDA, Washing-machine, Wrist-watch, TV.....
Anyway, it feels a bit like the difference between OO and Procedural coding, where OO projects tend to suffer without a reasonable amount of design effort *before* starting to code, but make maintenance simpler over the life of the project, and Functional-style tends to allow you to get stuck in, but can make maintenance and/or functionality upgrades more difficult. In fact, even down to the usual blaming of OO-style for creating slower code.
Horses for courses; but I'd love there to be a full-fledged Linux-style OS (FOSS, good driver support, etc) running on a MK. Just because;-)
I remember 'Plan 9' news around 10+ years ago.
It was distributed-only, where the Disk-subsystem ran as seperate (networked) nodes from the CPU-subsystem(s), which were seperate from the Terminal(s), etc, etc. It seemed an awful lot like a mainframe-style system using commodity parts, but you had to invest in at least 3 nodes in this way, if not more. This could have been expensive for what was mostly a research or hobby system at the time -- at least if you were going to get anything usable, speed-wise. (Others more intimate with the details may be able to correct/corroborate this)
Soon after, they released Inferno, which was basically the same idea, but the distributed-ness of the components was up to you.
I'd be surprised if much 3D software that old would still be considered anything like high-end.
I couldn't agree more about other (seemingly unrelated) skills having a big impact on your employability; I got my first real job in IT -- after only working in a PC Sales shop -- because I had a degree in psychology. I found out later that this was the main reason they employed me, and that they thought it would mean I had better skills in dealing with people in more diverse situations -- i.e manipulating them and their expecatations. Perhaps they were right, but it surprised me, as I would've said: without years of experience (either personal or commercial) or an 'innate' ability, it's just not the case!
I live in London, now, and it's 1 - 1.40 = NZ$2.80 - NZ$4 over here, so NZ$2 would be fine;-) But 1 is not too bad when you compare the average living costs over here, anyway.
And Red Bull (which is British, I believe) is no more than 20p cheaper; all the energy drinks are expensive, but I personally don't find many of them as chemical tasting as your average Coke, besides the fact that Coca-cola is nothing but industrial chemicals carefully (?) mixed... At least V has Guarana juice, even if it has plenty of other man-made chemicals in, as well.
I probably mis-remembered 'liver damage' for 'kidney damage', but paracetemol definitely has a damaging effect every time, and too-regular or too-big a dosage will begin to cause irreversible damage -- which is why the warning on the packet -- and codeine is even a restricted drug in the UK because of it, though strangely not in NZ or Australia.
I used to suffer from bad migraines, and it's unbelieveably effective against them where the others (incl. ibuprofen) won't make a dent. Pity, really, because my Grandfather being a doctor warned me of the side-effects.
I've been trying (though perhaps not too hard) to kick/moderate what I think's a caffeine addiction to an NZ energy drink called 'V'... except that it uses the juice of Guarana berries (from South America) as its source of caffeine.
According to a lot of various sources I've read, natural buffers and oils in the berry will slow the intake of caffeine drastically, which means a longer, slower 'high' and likewise a less drastic come-down (makes me sound like a catalogue).
My own 5-6 year's experience(s) of drinking the stuff seems to suggest its true, which is nice -- no headaches or cravings when I'm on holiday where it can't be bought -- but I still otherwise drink at least 2 cans a day, every day, and more if I don't limit myself...
But anyway, the point is I think it's a caffeine habit more than an addiction, except for the 'well-documented' addiction of caffeine, and the fact that some other addictions (heroine? alcohol?) have a strong psychologically-addictive factor such that being around addict friends or in places you used to get high can trigger a powerful craving...
It seems to me that caffeine is similar in that for some people it's more psychologically than physiologically addictive... which perhaps makes it a bit more insidious as addictions go...
You *must* be very careful with any pain-suppressant drugs. For one thing pain is the body's way of saying something's wrong, and masking it rather than fixing the problem is dangerous. But this is probably blatantly obvious when you're trying to give up caffeine;-)
More importantly, some pain-suppressants can cause other serious damage, i.e. paracetemol and codiene-based products *will* damage your liver, not only over time, but if you have too much in one go, you can hasten the damage.
For an example, I was told of a case in NZ last year (2003) of someone trying to O.D. on Panadol (paracetmol based). I wee bit ridiculous, as you can't do this in the way you might on barbituates, but after 100 or so Panadol pills, he quickly had enough in his system that he destroyed his liver and died slowly and painfully from blood poisoning (I think that's what happens after liver failure) over the next few weeks.
I don't think Aspirin can do this, but I do remember it being able to cause damage to some other organ almost as badly -- kidneys?
Slashdot Mirror
Funnily enough, I've been working on exactly this over the weekend. Unfortunately, the docs are terrible.
Some of the Wiki info suggests that things have improved, but I'm discovering a decent-sized client cache (10Gb) so that I can offline most of what I'd use has horrendous occaisional slow-downs and pauses.
I'm planning on testing a local/client server and a client with the RVM turned-off this week, but I'm not keen on the size that the RVM file(s) will have to get to.
Previous comments in here suggesting it's not really designed for modern data-sets (gigs rather than megs) are starting to look as if that's true... Other than that, it actually looks like a reasonably good design!
I'd like to second all of the above. I spent a long, long time trying and testing as many OSS SCM's as I could, and have decided that Subversion is "unfortunate". I like their multiple-architecure support -- using the apache cross-platform libs was rather inspired... but that's about it!
Being able to apply tags and branches on the exact same tree is conceptually more natural, even though Subversion's zero-copy/linking initially looks better.
I'm a big fan of darcs, too, and creating new repo's as branches is easy to get used to, but a little messy when you want to compare differences/changes.
In addition to this, the Subversions multiple full-body redesign has always made me very, very nervous. I still find it hard to believe that Subversion won't undergo another major redesign at any point!
To be honest, SVK almost convinced me of using Subversion, especially as it solved Subversion's fundamental difficulties with merging, but by then, Git and darcs had proved themselves far superior.
All the A.D. advantages you've mentioned are avail. under OSS, long before A.D. had them.
Perl modules for LDAP are old news. They're very useful.
In fact, I've made Perl update PDC data a lot more easily than anything else I've used.
You can easily create multiple hierarchies under one LDAP Base DN, and apply your ACLs based on the Base DN. You can even create multiple Base DNs running on different servers, and teach them how to pass clients off to each one.
If you really need strong auth and auth-domains, you should be looking at Kerberos. Even A.D. is based on it, and many, many products can use it (i.e. Samba, PAM libraries, etc).
I'm not too sure how it interacts with LDAP, but I believe it can with little pain -- in fact I vaguely remember OpenLDAP can auth. against Kerberos somehow...
And the thing about being able to use it "out of the box" implies a workable set-up as soon as it's installed, which just doesn't happen in the real world:
You read the docs/go on your course
You adjust your expectations
You install
You configure it properly for your environs
You test
You add some power-users
You fix
You get it signed-off
You deploy
(You fight off complaining users who are never satisfied)
Basically, if you're expecting to use A.D anywhere, you're really advised to stick to all-MS.
We worked hard on getting A.D. to play nicely with a Unix LDAP system, Bind (DNS), Samba, etc. and it just wasn't even slightly fun. There's quite a few hacks that they use, and they seem to expect an ability to dynamically-update quite a few things (e.g. in DNS) which was tricky to get going with Unix tools. On top of that, it will be expensive.
However, if you avoid A.D, and even Windows PDC's, it's actually fairly easy. OpenLDAP is mostly only tricky for Access-Controls, Samba 3 can do pretty-much everything SMB/CIFS file/print-related, and can auth. against LDAP easily.
We preferred Exim over Sendmail, Postfix, and QMail, but just pick the one you like best as they all do LDAP.
We installed Dovecot for the IMAP server -- does LDAP, too.
I think the main point is: if you use some decent (read: fully-compliant) LDAP server, or X.500 + LDAP shim, the rest of it can be whatever you like best.
I would like to put in a couple of other points:
However licenses start at £150-ish/user, and £3000-ish/server... (sorry if I mis-remembered those prices!)
Well, went there on holiday for the last fortnight, to see some friends who've been working there for almost a year.
My friends are very, very, luck, as their wages are paid by a British company, but they say most people there are unbelievably badly paid cf. Western countries, even when you take into account the difference in cost of living. e.g. a lot live in tin-shacks, just as you see in pictures on the TV of African shanty-towns, £60/month was a common wage. Also, import-duty is extremely high, too: 95% for imported cars, for example. So you don't see a great deal of "foreign conveniences".
Part of the "scam" might be the resorts are usually stunning there. The tourist packages will never show you the run-down areas. It reminds me of the Yucatan peninsula in Mexico (though perhaps less run-down in Mauritius) where the tourist resorts are lovely and the coastal towns are not a lot worse than some English ones, but when you hop in a car or take a tourist-coach through the inland towns you realise that people are only just surviving.
However, I'm told most Mauritians are much happier and friendlier cf. Londoners even so. The speculation there was the huge quantities of alcohol drunk by the general population -- when large portions of the country are covered in sugar-cane, it's hardly surprising there's a lot of cheap (and very excellent) rum.
NB: You'll almost never see a bottle of Bicardi there because it's just not as good as the local rum, and import prices make it more expensive, anyway.
But anyway, if you're expecting to be able to have a similar life-style to a modern Western city, you'll get a rude shock. If you're wanting to get away from the "modern-ness" of the Western cities, you'll probably enjoy it.
I'd heartily agree with learning a language for slightly different reasons; though programmers logically would probably benefit the most, the advantage is in having to translate concepts into a another form in a way that is less than automatic.
I personally don't think it matters hugely which language, as they *all* have idiosms (sp?) that you have to know how to deal with. e.g I learnt French in college and University, and it was a milestone to understand that certain phrases (some words, too, but that was more obvious) just can't translate properly, and the understanding had so much more to do with understanding the culture, traditions, history, etc than any of the words. I also learnt Maori (New Zealand) for short while, and found it bloody difficult until I stopped trying to translate any words independantly, and just worked with whole phrases.
These other effects on/of language are some of the things that I think are helpful with IC&T in general, partly because there is a strong culture and history there which dictates why some things are the way they are. e.g what does the phrase "religious war" mean to a Techie vs. non-Techie? e.g My wife would like someone to "just write a program" that'll make Hotmail work the way she wants without her having to remember the less-than-intuitive interface idiosyncracies; with her not knowing (or really wanting to know) all the detail that goes into Hotmail, the Internet, Client-Server, etc, etc, it's down to me to explain the relevant concepts in her "language" so she doesn't beat the computer into dust ;-)
As an aside: it seems to me the best programmers out there are those who can pick up real-world languages with some degree of ease.
If I remember rightly, it used to be you bought the hardware (at huge expense) and the software was thrown in, including source code. The hardware was mostly useless without it, after all.
I think you were forced to buy a maintenance contract and all that, but the software wasn't considered valuable enough in it's own right for a pretty long time. I don't think many companies copyrighted it or anything, either.
And isn't that how a lot of the old-school Unix hackers (read: "current OSS advocates") got going: hacking/fixing the vendor-supplied OS?
I've not tried s/w RAID-5 under Linux, as, among other things, a Solaris DSO course I went on recommended against it so heavily -- and not just because Sun wanted to sell us more hardware ;-)
But anyway, I've striped partitions across Linux (LVM2) and Solaris disks (Disk Suite/Vol Manager) several times -- RAID 0 I think that is -- and that seems to be pretty efficient.
Mostly it has to do with the fact that good h/w RAID has one channel and possibly bus dedicated per disk (whether SCSI or IDE) which means there's very low access-contention, but it also helps (a lot) that the dedicated controller handles all the splitting/consolidating of the data across the disks, parity-checks when a disk is replaced, etc, etc, so the OS can get on with what its better at.
Sharing a channel/bus (e.g a single IDE cable or a single PCI Bus) can really affect read/write performance because the data will have to be consolidated in a divided manner across those same channels/buses from all the disks (or all-but-one for RAID-5). The speed of those channels/buses can vary a lot between systems and architectures, though a 500 MHz CPU suggests a less-than-desirable speed for this task...
Parity-checking a replacement disk will also absolutely destroy disk-access speed -- *all* the data on *all* the disks has to be read in to write the correct parity to the new disk -- but a good h/w RAID should still allow the array to be used at the same time without too much pain.
Anyway, since you've got the disks already, it would be worthwhile playing with RAID and finding out for yourself, but if you can can't spare the time (or headaches) I'd definitely think seriously about a h/w RAID solution.
The 2 biggest reasons I gave on KDE, after trying it many times over the years, is that it still feels noticably slower, but more importantly: C++ completely locks me out of using some types of development.
C++ libraries on a C-Object design (think GObjects, etc) would probably allow me to use Qt easily and interact without some sort of middleware, while still abstracting the interface nicely.
On top of that, Qt's MOC/preprocessor style adds a further layer of cruft that I think is pretty unecessary these days.
I shouldn't knock it too hard, as it was a *good* idea at a time where the next best option was the Visual C++ on MFC style, but it's a harsh lock-in these days.
However, there does seem to be certain aspects of GNOME severely lacking. e.g. a simple Printer-widget/interface would make a big difference to me...
I personally think he has a good point about the Subversion design. VCS systems are pretty complicated beasts when it comes down to it, and deserve a fair bit of work on the design before doing any production-oriented coding.
From what I remember of Subversion, looking at it on-and-off over the years, they have ended up redesigning it several times over, while trying to produce code that would end up being part of the finished product, thus throwing away large portions several times because it didn't fit the new design, rather than the code could be done in a better way.
I also found it a little strange to be trying to basically implement a VC file-system with file-attributes... Why not actually make a VFS plug-in for Linux, even if it forces everyone to adopt a Linux Server for the repository, and do the merge, etc, tools at the user-level? It still seems more elegant that way, even if it forces use of of Linux -- didn't VMS have a rudimentary Rev Control in its file-system?
When I first saw Arch, I immediately shied away because it was all written in shell, however when the first version of 'tla' appeared (i.e the C-code version) I was quite impressed at the fact he was actually allowing himself to test his basic desing ideas in what amounts to a Rapid Dev Environment, even if Shell-script is a painful beast to try and do something like this in.
We have Debian here. We also have Solaris, SuSE (and a little bit of RedHat and Fedora for those who don't know any better) and used to have Gentoo.
We use Debian on any Linux servers -- primarily because Yast gets in the way for most serious Unix admins -- and SuSE on most workstations -- it makes it a lot easier for Noobs who actually need to be able to 'fiddle' with their box (driver development, etc). I like them both, but prefer Debian.
Our sales pitch is based on the fact that we usually wouldn't bother calling for vendor support, anyway. As fast as some of them are, they're usually not as fast as us researching and fixing it ourselves; you have the phone-call, then sometimes the Appeasment Engineer, then they want a config. dump so that they can look at the problem, then they often come back to you wanting more information... Whereas even with a bit of Googling and some mail-list trawling, someone's often described your problem and how they fixed it. There's not a huge number of 'new' problems out there.
On top of this, we started off with small introductions -- e.g a CUPS server was much easier and faster to get up-and-running on Debian than Solaris -- generally setting up the machine as proof-of-concept, then deciding the thing worked reliably enough not to migrate it elsewhere.
I think it ultimately doesn't matter who the vendor is, they're going to be rubbish at some point or for some thing, and if you're not going to be able to present yourself as the first and fastest trouble-shooter in a large percentage of situations, almost any OS/Distribution is going to bite you hard and often.
With Linux, I would say that percentage needs to be even higher, just because the tools and skills in use by the Support Engineers are probably going to be exactly the same ones available to yourself -- i.e any of the non-open-source vendor's proprietory tools and trade-secret knowledge is in the public-domain for open-source.
However, if you're mostly wanting to make it possible to redirect blame, then there's nothing better than a vendor.
I'm a huge fan of Ruby, and have been using it for years, however I still use Perl a lot, mostly because more people know how to use it at my work, and fewer people understand objective programming here, too.
But another reason I still use Perl is 'Perl 6', and the fact it seems to be incorporating a lot of the things I like about Ruby, plus a couple of other interesting ones, but also because the Parrot VM is supposed to allow you to write different parts in Perl or Ruby (or Python, etc?) or *whatever* is appropriate at the time.
I really like the idea of your HTML developer being able to use their polished PHP/HTML skills to suck the data through the nicely objectified RubyOracle libs your DB Developer wrote (instead of the rubbish method(s) PHP insists on now for accessing SQL-ised data...) which can also interact with the more up-to-date Perl6/LDAP libs, all because Parrot lets them work together at the code-level.
I'd never heard of Maporama before.
I'm from New Zealand, but left for Britain 5 years ago. In that time, this is the first mapping web-site that even knew NZ existed, let alone could map street names -> map locations...
Obviously I've just been using the wrong web-site, as I notice the maps are from NavTech, and I believe that's one of the more popular providers to Mapping web-sites.
Sorry, I should've added 'proof of purchase' as one of the requirements, but that's mostly to avoid shop-lifters getting refunds for things they've stolen.
As far as I understood it, they have to accept returns, so long as it's within a reasonable time-limit -- I think 28 days -- and in a 'resaleable' condition -- which usually means unopened and undamaged, and sometimes completely unused, e.g under-pants, but that can sometimes be too hard to prove.
Anything above that is at the shop's discretion.
I think it's also up to you to take it to court if they dispute it.
My information's mostly from TV programmes I've seen recently about kids being ripped-off because they didn't know these rights.
In Britain and New Zealand, a returns policy is mandated by law: provided the product is returned in a resaleable condition.
In fact, it's probably the same in Europe and the rest of Australasia.
Several stores in the UK will allow you to return CDs with copy-protection (i.e. the variety that causes Mac CD-drives to lock up, and won't play properly in most PC CD-drives) without any questions, too, because of the problems and confusion it's caused.
I've heard New Zealanders and Aussie's are popular for certain broadcast roles in the US because when they put-on a US accent, no-one can tell where they're from, therefore there's no regional bias to the speaker.
But as to the UK news-readers having a 'neutral British' accent, that's an utter fabrication. A lot of presenters have a slightly posh South-East or Midlands accent, but it's just as likely you'll hear Northern, Irish, Scottish, Cornish, Welsh, etc, etc. It's more that no-one really cares much about the presenter's accent, only that they come across as appropriate to that TV show's audience, and that they can be understood.
As an aside: I've never heard anyone able to put-on a proper Kiwi accent who hadn't lived there for at least 2 years. It's easy to recognise an Aussie one, for example, even a Western-Australian one, if you've lived in NZ for long enough, and I'd say Aus and NZ accents are very close cousins.
I'm sure this point's been argued elsewhere, but: the whole reason for regulation and legislation over any and all employment discrimination is because the companies would otherwise always have the upper-hand in any employee-related dispute, and most companies do not act in anything like a loyal manner to their employees. This is not 'we can no longer support this position', which is a result of change and competition -- the employee also has the right to say 'I can no longer work in this position -- but more along the lines of, 'we don't care how much you've given in time, effort and expertise, while you have this condition or disability, because our accountants think you're statistically too much of a liability'. Statistics and lies, no common sense.
So many people have bought into a management style that emphasises live to work, even if it kills you, and this esoteric and unrealistic concept of 'growth,' year-on-year, forever, that there is becoming more of an insistence on employees which fit into some kind-of super-human species, or the nearest to that available, that we're ending up with stressed-out, over-tired, and genuinely worried employees instead.
The idea that a young employee will be any better than an older one seems to be based on the enthusiasm of youth -- which jades pretty quickly -- and ignores its stupidity and self-centredness.
The idea that a pregnant woman being 'off sick' for 6 months id too much of a liability is very short-sighted. I've heard excuses that they'll decide to stay at home for 5 years looking after kids as a housewife, anyway, but it's no longer anywhere near as accurate as 30 years ago, esp. to the modern career-woman.
The idea that your culture and its stereotype/reputation will be an accurate indication of work ethics for every individual is frequently proved wrong.
And all this is supported by legislation and regulation because of the simple fact that people are too-often short-sighted, management too ruthless/over-worked/incompetent, and companies too focused on competition, profit and 'growth'.
I am glad the government tries to support the people as a whole more than minorities and individuals -- incl. in the previous post's 'company' comparison to indivuals -- at the same time as trying to help minorities and individuals to survive and/or thrive. It's a balancing act that could so easily fall to bits, and despite the modern capitalist's support of letting only the strong and unemotional survive, at the expense of our humanity and neighbours, it'd be a terrible, Klingon-esque world if we didn't keep working on that balance.
I personally think Microkernels are *still* a good idea. A well-designed one is extremely modular, often extremely small (yeah, I know: "duh") and very portable, and the small-chunk approach can make them a lot easier to understand. On top of this, you can seperate the different parts of the kernel from each other, so there's far less likelihood of different pieces stepping on others, drivers can (theoretically) be run in user-space (if such a thing applies)...
;-)
However, like anything, they are *not* going to be a perfect panacea. If the design is not just right, they can end up being far worse than the mono-kernels they are often competing with.
The Mach MK is one where kernel-developers complain frequently about trying to get drivers to work with it, its message-passing architecture being unable/difficult to use in certain important ways, etc, etc, whereas I've heard a lot of praise for the L4 design.
In fact, isn't TRON/iTRON (etc) a MK-style design? If it weren't for the US Trade Dept, it'd probably be running on your Desktop, PDA, Washing-machine, Wrist-watch, TV.....
Anyway, it feels a bit like the difference between OO and Procedural coding, where OO projects tend to suffer without a reasonable amount of design effort *before* starting to code, but make maintenance simpler over the life of the project, and Functional-style tends to allow you to get stuck in, but can make maintenance and/or functionality upgrades more difficult. In fact, even down to the usual blaming of OO-style for creating slower code.
Horses for courses; but I'd love there to be a full-fledged Linux-style OS (FOSS, good driver support, etc) running on a MK. Just because
I remember 'Plan 9' news around 10+ years ago.
It was distributed-only, where the Disk-subsystem ran as seperate (networked) nodes from the CPU-subsystem(s), which were seperate from the Terminal(s), etc, etc. It seemed an awful lot like a mainframe-style system using commodity parts, but you had to invest in at least 3 nodes in this way, if not more. This could have been expensive for what was mostly a research or hobby system at the time -- at least if you were going to get anything usable, speed-wise. (Others more intimate with the details may be able to correct/corroborate this)
Soon after, they released Inferno, which was basically the same idea, but the distributed-ness of the components was up to you.
I'd be surprised if much 3D software that old would still be considered anything like high-end.
I couldn't agree more about other (seemingly unrelated) skills having a big impact on your employability; I got my first real job in IT -- after only working in a PC Sales shop -- because I had a degree in psychology. I found out later that this was the main reason they employed me, and that they thought it would mean I had better skills in dealing with people in more diverse situations -- i.e manipulating them and their expecatations. Perhaps they were right, but it surprised me, as I would've said: without years of experience (either personal or commercial) or an 'innate' ability, it's just not the case!
I live in London, now, and it's 1 - 1.40 = NZ$2.80 - NZ$4 over here, so NZ$2 would be fine ;-)
But 1 is not too bad when you compare the average living costs over here, anyway.
And Red Bull (which is British, I believe) is no more than 20p cheaper; all the energy drinks are expensive, but I personally don't find many of them as chemical tasting as your average Coke, besides the fact that Coca-cola is nothing but industrial chemicals carefully (?) mixed... At least V has Guarana juice, even if it has plenty of other man-made chemicals in, as well.
I probably mis-remembered 'liver damage' for 'kidney damage', but paracetemol definitely has a damaging effect every time, and too-regular or too-big a dosage will begin to cause irreversible damage -- which is why the warning on the packet -- and codeine is even a restricted drug in the UK because of it, though strangely not in NZ or Australia.
I used to suffer from bad migraines, and it's unbelieveably effective against them where the others (incl. ibuprofen) won't make a dent. Pity, really, because my Grandfather being a doctor warned me of the side-effects.
I've been trying (though perhaps not too hard) to kick/moderate what I think's a caffeine addiction to an NZ energy drink called 'V'... except that it uses the juice of Guarana berries (from South America) as its source of caffeine.
According to a lot of various sources I've read, natural buffers and oils in the berry will slow the intake of caffeine drastically, which means a longer, slower 'high' and likewise a less drastic come-down (makes me sound like a catalogue).
My own 5-6 year's experience(s) of drinking the stuff seems to suggest its true, which is nice -- no headaches or cravings when I'm on holiday where it can't be bought -- but I still otherwise drink at least 2 cans a day, every day, and more if I don't limit myself...
But anyway, the point is I think it's a caffeine habit more than an addiction, except for the 'well-documented' addiction of caffeine, and the fact that some other addictions (heroine? alcohol?) have a strong psychologically-addictive factor such that being around addict friends or in places you used to get high can trigger a powerful craving...
It seems to me that caffeine is similar in that for some people it's more psychologically than physiologically addictive... which perhaps makes it a bit more insidious as addictions go...
You *must* be very careful with any pain-suppressant drugs. For one thing pain is the body's way of saying something's wrong, and masking it rather than fixing the problem is dangerous. But this is probably blatantly obvious when you're trying to give up caffeine ;-)
More importantly, some pain-suppressants can cause other serious damage, i.e. paracetemol and codiene-based products *will* damage your liver, not only over time, but if you have too much in one go, you can hasten the damage.
For an example, I was told of a case in NZ last year (2003) of someone trying to O.D. on Panadol (paracetmol based). I wee bit ridiculous, as you can't do this in the way you might on barbituates, but after 100 or so Panadol pills, he quickly had enough in his system that he destroyed his liver and died slowly and painfully from blood poisoning (I think that's what happens after liver failure) over the next few weeks.
I don't think Aspirin can do this, but I do remember it being able to cause damage to some other organ almost as badly -- kidneys?