*hard* - at least I understand the problem, but curses on the restrictions of a binary tree =)... no idea on algorithm that would finish executing before the end of the universe.
I only scanned the article and problem descriptions, but isn't the final problem a similar idea to Huffman coding?
Actually, although I'm a fan of C++ generally, I think they got this one wrong. Using << and >> in this way encourages you to put the format of a composite string in code and not in data. That immediately creates problems with maintenance, flexibility and portability.
This is particularly damaging in the area of i18n. For example, you can give the string
Put the $1 onto the $2.
to a translation agency, and get back the translation into the language of your choice, with the order of $1 and $2 swapped around if the new language requires it. You can't do that if your string is all broken up in code.
I don't like printf much in terms of clarity -- it's very "old skool hackery" in that you have to know how it works before it makes any sense, and you still forget details -- but at least it put the logically whole unit, the string template, in one place. You can systematically construct a printf format string using any other string tools, you can send it off for translation as a unit, it's hard to introduce inadvertent changes during later edits, etc.
The ability to extend the IOStreams system to allow I/O of new types and to allow I/O of existing types to new streams is great, but we could certainly do better now building on that experience, by keeping the extensibility without giving up the rest.
Types where op overloading can provide a natural, intuitive notation that fits with the rest of the langauge? Off the top of my head, op overloading can provide:
natural notation for any mathematical type: Complex, BigInteger, FixedPoint, Matrix,...
the usual [] indexing notation for any suitable container type: Array, Hash,...
accepted notation for any textual type, in particular + for concatenation: Char, String, Regex,...
Complex numbers may not be common, but mathematical types, indexed containers and textual types generally sure are.
Because adding operator overloading to a language is like giving the programmer the ability to change and mutate the language to her desires.
No. It lets you extend the language's operator systems so that the other extensions you create by adding new types can work the same way as what's already there. This is invaluable if your language also supports generic programming, and aids readability generally.
It doesn't let you change anything in a sane language. You can't overload operator+(int, double) in C++, for example, because any overloaded operator must use at least one UDT.
Initial driver/game stability problems were bad with the 9700Pro, but they do seem to have gotten their act together now, particularly the last few weeks now we've got the Catalyst 3.2 drivers. As ever, the usual "buy at least six months late" advice was sound -- so rather than going for a 9800Pro and picking up the inevitable stability and performance issues all over again...
Back at the start of the year, I was deciding between a Radeon 9700 Pro and waiting for the new GeforceFX card to go in my new PC. After doing some homework, I opted for the 9700 Pro, but I was hearing from several boards (including this one, IIRC) that I was being a fool, the nVidia card would be out Any Day Now(TM), the ATI drivers sucked, the hardware would be abandoned, nVidia drivers would get big performance gains later on, etc.
Nearly six months after I started hearing this, we finally (several months late) have that nVidia competitor card. Essentially by their own admission, it is not strong in comparison to ATI's kit.
In those six months, ATI have released updated Catalyst drivers regularly, and now seem to have resolved most of the stability issues. And guess what? Their performance is up considerably with the 3.2 drivers, too.
I dislike lack of service, and if ATI abandon me the moment they move to the next hardware series, they will lose me as a customer for a goodly long time and I will tell everyone I know to avoid them, and why. But I also give credit where it's due, and solid, demonstrable hardware and drivers from ATI will get my money way faster than hypothetical anything from nVidia fans any day.
'Course, I shoulda bought a 9500Pro and O/C'd, but you live and learn.:o)
Or you could buy a 9500Pro or 9700Pro now, depending on your budget, and save a fortune. Either card is still very good in today's market, and certainly the 9700Pro still has more features available than any current game is even near to using. I bought mine too early (start of the year), but now they're cheaper and the most recent Catalyst 3.2 drivers seem to be a significant improvement for both performance and stability on the above cards. I'd take either over a shiny new 9800(Pro) or nVidia kit today.
So introducing a feature lean product that focuses on the fundamental features that users actually use on a dialy basis is what I think project like Open Office are or should be striving for.
I totally agree that few people use or care about advanced macro facilities, setting up whole document production lines, etc. as supported by MS Office apps. However, mail merge, document templates, maybe some of the group working facilities etc. are important.
The thing is, though, the usability points are mostly in these major features anyway. OpenOffice's styles are great in principle, but broken in several serious ways in practice: you apparently can't use the "inheritance" idea because the UI just isn't there, and I still haven't worked out whether it's possible to assign shortcut keys to styles (rather than reaching for the mouse literally every word sometimes) and whether there's a shortcut to remove all character formatting and revert to the paragraph default. The mail merge has the same sort of problem: big bugs make it very awkward to update a mailing list you're keeping in a spreadsheet, for example. The version I got (1.0.1 as downloaded from their web site) doesn't even have a UK English spelling checker, nor was there any obvious way to get one when I was downloading it. These things are showstoppers for a business that wants to install a standard set-up and have their users ready to go, and the reason I say MS Office still way ahead on usability.
I actually work on the Visual Studio team, and I am VERY interested in specific gripes and suggestions for improvement. Can you let me know what you most like and dislike?
Obviously I don't know who you are, but I have no reason not to take you at your word, so here goes. For background, I have been the "guinea pig" for a move to VS.NET at two different offices within the past two years, in different fields but both in software development with around 30-40 developers, pretty much all using VS. I don't hold the purse strings, but I get asked for an opinion by those who do. Our main interest is in C++ and supporting tools. We don't much need the GUI stuff and wizardry (sometimes, but not often). We really couldn't care less about.NET.
OK, so first, to give credit where it's due, some good points.
Standards compliance is much better in recent versions of VC++.
I introduced my new office to AUTOEXP.DAT and it caught on real fast. It's way more valuable to us than almost every bullet point on the average marketing feature list for the VS debugger.
Little-seeming things like those line numbers you can now display at the left of an editing window sometimes really help.
And now, in the hope that this will make a small difference somewhere, my pet gripes -- me and half the Windows development world, I'm guessing.:-)
You took away the old browse toolbar in VS.NET. You replaced it with something so cumbersome and underpowered that I can't even remember what you call it. Please bring it back, or provide something even better.
You have tabs now (good) but WndTabs doesn't work (very, very bad). I assume you're all very familiar with the latter tool; if not, it's homepage is here. It's better than yours by a long way. Don't just give Oz a freebie copy of VS.NET to develop with, contract him to write the support right into VS, or get one of your own teams to do it. Whatever you have to pay to get it done is going to be worth it.
In the solution browser in VS.NET, the tree now seems to home in on the file you've currently got open. That's irritating at best. The fact that the project properties option (where things like command line parameters for a debug run are set) disappears from the main menu as well unless the project is highlighted in the browser is just downright infuriating.
None of these things is rocket science on a project like Visual Studio, and two of them (first and third) were there and worked fine in VS6 and have been broken for no apparent reason since. I promise you quite honestly that these issues are directly costing you thousands of pounds in lost sales in the UK at my offices alone, because when people hear about them while asking me whether to upgrade, they are losing interest and sticking with VS6.
I'd be quite happy to elaborate on these points and/or provide further ones if you want. I don't know how much one man's opinion means to you, but certainly your main marketing is telling me all the wrong things at the moment.:-( So, I'm guessing it's in both of our interests for you to know what I think. Here probably isn't the right place, though, so post some sort of contact e-mail address if you like and I'll get in touch.
OpenOffice provides everything that a significant percentage of office workers need to get their job done.
I honestly disagree with that, for the reasons I've detailed in another post to this thread. For individual use, it's fine. But it's too unstable dealing with MS Office documents, too buggy in areas commonly used by businesses and too far behind in the usability race. The cost of entry is just too high today for most businesses. Give it a couple of years and I think it will be a very real contender, but not yet.
there is NO reason to buy M$ now. OSS OpenOffice is great NOW. I use OpenOffice almost every day. More on point - it is the *only* word processor/suite I use after migrating from StarOffice. There is NO reason to choose (buy) Word over OpenOffice.
As I've mentioned elsewhere in this thread, I also use OpenOffice regularly, on my home PC. I have worked extensively with the word processor and spreadsheet apps, and have given the graphics app a serious try.
And I'm sorry, but I still completely disagree with you about its suitability for businesses. Among the good, and indisputable, reasons to remain with Microsoft Office for now are...
OpenOffice is not reliable importing or exporting MS Office files. It can be crashed importing very simple MS Word documents, and exporting back to.doc format gets far too confused to be acceptable in a business environment. Guess who's been trying to transfer some documents just this week...
Microsoft Office is still quite some way ahead on usability. (Did you see my other post in the thread where I cited several specific examples?) For geeks, this doesn't matter so much. In a business environment, where people just don't have the time or inclination to play around and get something right, it's not good enough.
OpenOffice simply has too many bugs in core features. Mail merge data sources are screwed when you're trying to combine form letters in the word processor with data sets in a spreadsheet. The tools for setting up proper document templates and styles are hit-and-miss in several places. The graphics tool has severe bugs with output quality when printing. These things matter in a business environment.
If OpenOffice works for you and does what you need it to, great, I have no problem with your saying that. But please don't go claiming it's ready for everyday business desktop use when it's so easy to point out things like the above. Potential? Yes, it's going to be fantastic one day. Current use? No way, for most office environments. And the latter is what this thread is about.
Right. Making victimless, self-abuse a crime results in a cost to society, because now there are more crimes and more activities where one must associate with criminals if one wishes to participate in those activities. Your point?
No, these crimes result in a much more practical impact: they cost health and counselling services a fortune to look after those who get themselves hurt and then need help to recover.
I'm not personally a drug user, but as far as I can see, cannabis is widely used and has far less of a health implication than alcohol or tobacco. There is a very good argument for legalising it, not least that there's no particular reason not to and most people in many countries where it is currently illegal seem to be in favour of the change.
What I'm talking about here is the hard drugs -- the ones that kill, or do serious damage, without a doubt -- which are used by far fewer people, but cause the vast majority of the mess (as well as the links to other organised crime, yada yada).
Prostitution is a whole different thing, but the simple fact is that most of the people involved in that line of work aren't doing it out of choice. They're abused, tricked, blackmailed, feeding drug habits, or otherwise motivated by something other than their personal desire to shag lots for money.
If you have to resort to breaking the law in order to catch the law-breakers, then there's something wrong with the system.
In cases such as those you cite, there are generally specific legal exemptions for those taking part in a law enforcement capacity.
If there were not, everything from high speed police pursuits to arrest using firearms would be illegal. Do you really think that's a good idea?
My issue is more with victimless crimes like prostitution and drug use,
If only that were so. Alas, it's often the case that people involved in these things, while perhaps (though certainly not always) starting the activity voluntarily, don't continue it out of choice. Moreover, the cost is often to society as a whole, when it has to clear up the resulting mess.
Again, copyright only covers the act of copying and distribution, not possession.
In the UK, it appears to cover the act of copying, full stop. There are specific exemptions and things aren't always black and white, but I'm not aware of anything that requires you to distribute a copy before your actions become illegal. Then again, I'm not a lawyer, I only deal with these issues in relation to some of my hobbies, so please cite anything that proves me wrong; you'd make my life much easier and save my friends and I a lot of money!
In most countries, it is not illegal to download copyrighted music.
If you're going to make claims like that, do please state which "most countries" you're talking about, and whether or not you are a lawyer or otherwise qualified to offer that opinion. What you say is certainly not the case in many western jurisdictions, and with no disrespect intended, your arguments read like a person who's wishing the law worked his way, not an informed legal opinion. I'd love to see you cite test cases in any western jurisdiction and prove me wrong, but I doubt that's possible.
Um... No. What you just wrote is technically infeasible with any of the currently mooted DRM technologies, and there is no legal basis, current or proposed, to support it, or to defend the perps who have "accidents" from accusations of malicious activity. You are spouting panic-induced FUD, and you desperately need to do some homework on what is, and is not, part of the current DRM ideas, both technical and legal.
Microsoft development tools (i.e., those sold by Microsoft) are, contrary to popular religion, not the cat's hind end. I'm a developer and I've found that Visual Studio.NET and Source Safe and nearly every other MS "product" I try is the most ridiculously bloated and cumbersome tool imaginable. I really can't understand how people use them to do real work.
SourceSafe is a fairly horrible product, it's true, but I guess we'll just have to agree to disagree about Visual Studio. I'm also a professional developer, and have been a "guinea pig" for a move to a new version of Visual Studio twice in recent years.
At first, I didn't like the latest versions. They have their flaws, they certainly could be less bloated, and Visual Studio.NET mysteriously removed a couple of very nice features that were present in VS6. But even ignoring all the.NET hype (which is irrelevant to me since my code runs on 15 different platforms), after keeping an open mind and giving it a chance, I find VS.NET to be a nice development environment. The recently released VS.NET 2003 seems to fix some of the more glaring issue with the original VS.NET, too.
VS.NET is very configurable. It can do all the same things with make files and such as the alternatives you mentioned, and has enough compiler options to make even GCC blush. The editor is pretty decent. The compiler outputs very high quality code compared to the field (and I work in mathematical software, so things like optimisations and generally efficient output code matter here).
I think the top spot for me is the debugger. A lot of my work involves tracking down bugs in complex numerical algorithms, and I make regular use of many of the VS debugger's more advanced features. An AC replied to me suggesting things like Qt Designer. Sure, some people have nicer GUI designers out there, but I'm talking about serious tools, not eye candy generators. What is the best debugger available in the Linux/GCC/LIDE world?
I'd love to know where you got your ideas about what's taught in university these days from, BTW. I'm not in academia myself any more, but several of my friends are in CS research at major UK universities, so I'm pretty well up on the academic scene here. I'm not aware of any group that specifically teaches anything much about the tools you mentioned; most of the universities, for better or worse, seem to use Java as their main teaching language now, and for the scale of project we're talking about there, the tools you mentioned are serious overkill.
You would be zippy to if you were loaded at startup and never taken out of memory, and had on OS desinged around you.
Gee, what are those little icons in my system notifications area?
Oh yeah, it's the quickstarters for Mozilla and OpenOffice.
But seriously, we're talking complete packages here. Microsoft do provide an OS and Office combination that work well together. Whether you like their commercial practices or not, businesses can buy into that and gain its advantages today. If Linux/OpenOffice/whatever can't match it, that's their problem, not Microsoft's.
I just love people like you. Just like my two year old, always in the moment and can never think of the future. OpenOffice.org is great now, now super, not fantastic, just great. And it will get BETTER.
I just love people like you. In a discussion about an article that questions whether there is any reason to buy Microsoft now rather than open source, you're looking at where open source might go later.
The question at hand is whether or not there's still any compelling reason to buy MS software today. The answer, for businesses looking for office software, is yes, for the reasons I and others have given in this thread. We all know open source has a lot of potential, but what happens tomorrow is a different question, and you don't know the answer to that any more than I do.
The biggest difference between those and the Microsoft equivalents is (1) load time [especially on windows] and CPU-hogging.
On the contrary; I find OpenOffice to be quite zippy on my WinXP box, at least comparable to MS Office.
The places OpenOffice is still playing catch-up to MS Office are much more serious at the moment: stability, feature set and ease of use.
I mainly use the word processor, for a variety of different document types from simple letters to DTPing programmes for events to technical reports. I also use the spreadsheet, both for numerical work and as a toy database and mail merge data source.
I have found that both apps are generally less resilient than the MS Office equivalents, particularly in the much-advertised import and export of the de facto standard Microsoft file formats.
There are some features that apparently aren't properly or completed supported, at least not yet, though for most everyday uses either app is adequate for most people.
Perhaps the biggest issue is all the "little things". MS Office has had a lot of usability research done on it. You may not like the paperclip, adapting menus or wiggly spellchecker underlines, but you can switch them off. However, to pick on styles as a usability example, OpenOffice has a nice theoretical mechanism, but where are my keyboard shortcuts for styles and special characters in OpenOffice? Where's the key to reset the character format back to the paragraph style's default? How do I reset a paragraph style to not specify any font changes from the base style? There is functionality approximating most of these things, but it just isn't quite tidy somehow. You can cite similar problems with many other specific areas of the code.
It's got great potential, but one good answer to the question "Any reason to buy Microsoft?" is simply yes, because (for now at least) it's still quite a way ahead on usability.
Yeah, because LOTS of enterprises run Office on their servers, right? Please RTFA.
If you had R'd TFA then you'd spot that a lot of it does relate to things other than the OS side of the market, and many of the claims made are general and across the board. In fact, from the original article, and citing a guy from SuSE of all places:
"The concern is the user's experience," Migliaccio noted. "The business user doesn't know much about the operating system or interact with it. The question is, do [the applications] provide the functionality they need, and can [IT] support them?"
Some of the other more telling quotes from the article follow.
Five years ago, the answer would have been easy. With the dominant development tools, client operating system and client applications, Microsoft owned a certain portion of the enterprise
That is still true, if anything more so today than it was five years ago. MS still totally dominate the desktop. In particular, their Windows development tools and office suite still completely outclass the OS equivalents. To give credit where it's due, a couple more years at this rate and OpenOffice could be a real threat. I haven't seen any open source project currently in development that's even close to Visual Studio.
Almost everything enterprises once found unique to Microsoft they can now find somewhere else -- without some of the baggage that comes with Microsoft purchases, like ongoing security concerns and mystifying licensing practices.
Because of course, open source things are immune to bugs and security problems... not. If you really think "almost everything" that was once unique to Microsoft now has a serious open source competitor, you haven't been looking very carefully.
On the desktop side, Linux is also providing viable alternatives to Windows. For example, Linux vendor SuSE recently introduced Office Desktop, a Linux product that includes a copy of Sun StarOffice 6.0, which is a competitor to the Microsoft Office suite.
It may be a competitor, but it ain't a better product. It's got a way to go before it challenges either the raw power or the ease of use of the Microsoft suite. For geeks who are happy to play with new toys, it's great, and maybe in the future it will be great for Joe Average as well, but enough with the kidding ourselves, OK? It isn't there yet.
The problem is that some syntatic sugar doesn't actually increase readability. Consider operator overloading.
OK. I claim that operator overloading can be used sensibly, and that contrary to claims of its detractors, this seems to happen most of the time even in C++. I further claim that when used sensibly, operator overloading significantly increases readability and maintainability. It is also a necessary feature to co-ordinate with writing generic algorithms using tools like C++'s templates. (Yeah, they have nasty syntax at times, but they're way better than nothing.)
Let's say y is a char* and z is an integer. Normally then the expression (y+z) would mean in C++ "a char pointer which points at the character at index 'z' of the string stored in y." But with operator overloading in the language, you can never be too sure.
Of course you can be sure. You can't overload operator+(char*, int) since it takes two built-in types and overloaded operators must use at least one UDT. You can overload operator&&, which in my view is a flaw in the current C++ scheme, but this is a well-known issue and standard practice is not to do it.
If you really think operator overloading is a bad thing, perhaps we should all give up forming strings using "Hello, " + name + ", how are you?" and go back to good ol' strcat and friends? Of course, Perlesque interpolation ("Hello, {name}, how are you?") is often cleaner still in this particular case, but one thing at a time!:-) (And it has its own disadvantages too.)
In general, the overloading of arithmetic operators for numeric classes such as matrices or complex numbers, the overloading of indexing operators for container classes, the overloading of increment/decrement/dereferencing operators for iterators and such all lead to code that looks more consistent than if you have a mishmash of arbitrarily named functions for each type to do the same thing.
As long as the person who writes the overloaded operators does so sensibly -- and you can write any function to have stupid side-effects, so this is no different to writing any other code -- the benefits in consistency are valuable aids to readability and maintainability. Don't look at what can be in the worst case, look at what can be in the best case, and most important of all, what actually happens. In this case, history amply demonstrates that operator overloading can be of genuine value in real world projects.
I'm in a similar position. I work for a company where I guess the average developer is 25-30, the average team leader 30-35 and there are several significantly older guys around. We have few managers; the older guys still write code like the rest of us for most of their time.
I saw a set of figures once (in the Mythical Man Month or something similar? Can't recall...) for the productivity of a developer. Basically, it claimed that over-30s could successfully work on more than one project at once. Over-25s were OK on one project, but not good at multitasking. Under-25s were pretty much only useful for documentation; give them a serious development role to play and they do more harm than good.
While I don't entirely agree with this from personal experience, I can well believe that taken across the whole industry these figures are a reasonable estimate. Certainly a lot of just-graduated kids think they're clever and dive in trying the latest and greatest tools, and promptly make a huge mess that the older and wiser then have to clear up (while still doing their own stuff as well, of course). Meanwhile the more mature and experienced developers are much more sceptical of new shiny things, and prefer tried and tested tools and techniques until new ideas have been proven in practice. That in itself makes them much less of a liability than the go-faster-stripes generation.
There is a natural problem with many industries, software development included, that the Peter Principle applies. Coders are promoted to their level of incompetence. Often they're also transferred out of development and into management, not because they're going to be more useful there or have shown any particular aptitude for management, but simply because they're older and the existing stupid managers think that doing this will help.
I've noticed in my time in the industry that where you have people who are still close-to-the-metal developers in their late 30s -- by which time many have dropped into the management pool -- and beyond, it's usually because they've made an active decision to remain that way, and they're usually the kind of people you would want to work with as a developer, or to whom you could entrust your vital project as a manager. These people should be given the respect they deserve and treated like the gold dust that they are, not sidelined because their DoB is earlier than 1980. Most of the more successful businesses I've seen, particularly in hard times such as the past couple of years, seem to take older developers seriously. Coincidence? I think not.
But what would I know? I'm only 25, and just about able to work on a single development project without destroying it.;-)
Its embarrassing to even admit that I've used [VB] before though.
Why? Although I've never used it "seriously" myself, even I can see its obvious attraction for some types of job, and that it would be more appropriate than C++, Java, Python, or whatever other alternative. As with anything, you should pick the right tool for the job, whether or not it happens to be from Microsoft and whatever its reputation amongst certain types of programmer.
Slashdot Mirror
I only scanned the article and problem descriptions, but isn't the final problem a similar idea to Huffman coding?
Actually, although I'm a fan of C++ generally, I think they got this one wrong. Using << and >> in this way encourages you to put the format of a composite string in code and not in data. That immediately creates problems with maintenance, flexibility and portability.
This is particularly damaging in the area of i18n. For example, you can give the string
to a translation agency, and get back the translation into the language of your choice, with the order of $1 and $2 swapped around if the new language requires it. You can't do that if your string is all broken up in code.
I don't like printf much in terms of clarity -- it's very "old skool hackery" in that you have to know how it works before it makes any sense, and you still forget details -- but at least it put the logically whole unit, the string template, in one place. You can systematically construct a printf format string using any other string tools, you can send it off for translation as a unit, it's hard to introduce inadvertent changes during later edits, etc.
The ability to extend the IOStreams system to allow I/O of new types and to allow I/O of existing types to new streams is great, but we could certainly do better now building on that experience, by keeping the extensibility without giving up the rest.
Types where op overloading can provide a natural, intuitive notation that fits with the rest of the langauge? Off the top of my head, op overloading can provide:
Complex numbers may not be common, but mathematical types, indexed containers and textual types generally sure are.
No. It lets you extend the language's operator systems so that the other extensions you create by adding new types can work the same way as what's already there. This is invaluable if your language also supports generic programming, and aids readability generally.
It doesn't let you change anything in a sane language. You can't overload operator+(int, double) in C++, for example, because any overloaded operator must use at least one UDT.
Initial driver/game stability problems were bad with the 9700Pro, but they do seem to have gotten their act together now, particularly the last few weeks now we've got the Catalyst 3.2 drivers. As ever, the usual "buy at least six months late" advice was sound -- so rather than going for a 9800Pro and picking up the inevitable stability and performance issues all over again...
Back at the start of the year, I was deciding between a Radeon 9700 Pro and waiting for the new GeforceFX card to go in my new PC. After doing some homework, I opted for the 9700 Pro, but I was hearing from several boards (including this one, IIRC) that I was being a fool, the nVidia card would be out Any Day Now(TM), the ATI drivers sucked, the hardware would be abandoned, nVidia drivers would get big performance gains later on, etc.
Nearly six months after I started hearing this, we finally (several months late) have that nVidia competitor card. Essentially by their own admission, it is not strong in comparison to ATI's kit.
In those six months, ATI have released updated Catalyst drivers regularly, and now seem to have resolved most of the stability issues. And guess what? Their performance is up considerably with the 3.2 drivers, too.
I dislike lack of service, and if ATI abandon me the moment they move to the next hardware series, they will lose me as a customer for a goodly long time and I will tell everyone I know to avoid them, and why. But I also give credit where it's due, and solid, demonstrable hardware and drivers from ATI will get my money way faster than hypothetical anything from nVidia fans any day.
'Course, I shoulda bought a 9500Pro and O/C'd, but you live and learn. :o)
Or you could buy a 9500Pro or 9700Pro now, depending on your budget, and save a fortune. Either card is still very good in today's market, and certainly the 9700Pro still has more features available than any current game is even near to using. I bought mine too early (start of the year), but now they're cheaper and the most recent Catalyst 3.2 drivers seem to be a significant improvement for both performance and stability on the above cards. I'd take either over a shiny new 9800(Pro) or nVidia kit today.
I totally agree that few people use or care about advanced macro facilities, setting up whole document production lines, etc. as supported by MS Office apps. However, mail merge, document templates, maybe some of the group working facilities etc. are important.
The thing is, though, the usability points are mostly in these major features anyway. OpenOffice's styles are great in principle, but broken in several serious ways in practice: you apparently can't use the "inheritance" idea because the UI just isn't there, and I still haven't worked out whether it's possible to assign shortcut keys to styles (rather than reaching for the mouse literally every word sometimes) and whether there's a shortcut to remove all character formatting and revert to the paragraph default. The mail merge has the same sort of problem: big bugs make it very awkward to update a mailing list you're keeping in a spreadsheet, for example. The version I got (1.0.1 as downloaded from their web site) doesn't even have a UK English spelling checker, nor was there any obvious way to get one when I was downloading it. These things are showstoppers for a business that wants to install a standard set-up and have their users ready to go, and the reason I say MS Office still way ahead on usability.
Obviously I don't know who you are, but I have no reason not to take you at your word, so here goes. For background, I have been the "guinea pig" for a move to VS.NET at two different offices within the past two years, in different fields but both in software development with around 30-40 developers, pretty much all using VS. I don't hold the purse strings, but I get asked for an opinion by those who do. Our main interest is in C++ and supporting tools. We don't much need the GUI stuff and wizardry (sometimes, but not often). We really couldn't care less about .NET.
OK, so first, to give credit where it's due, some good points.
And now, in the hope that this will make a small difference somewhere, my pet gripes -- me and half the Windows development world, I'm guessing. :-)
None of these things is rocket science on a project like Visual Studio, and two of them (first and third) were there and worked fine in VS6 and have been broken for no apparent reason since. I promise you quite honestly that these issues are directly costing you thousands of pounds in lost sales in the UK at my offices alone, because when people hear about them while asking me whether to upgrade, they are losing interest and sticking with VS6.
I'd be quite happy to elaborate on these points and/or provide further ones if you want. I don't know how much one man's opinion means to you, but certainly your main marketing is telling me all the wrong things at the moment. :-( So, I'm guessing it's in both of our interests for you to know what I think. Here probably isn't the right place, though, so post some sort of contact e-mail address if you like and I'll get in touch.
I honestly disagree with that, for the reasons I've detailed in another post to this thread. For individual use, it's fine. But it's too unstable dealing with MS Office documents, too buggy in areas commonly used by businesses and too far behind in the usability race. The cost of entry is just too high today for most businesses. Give it a couple of years and I think it will be a very real contender, but not yet.
Yes, you do. It's trivial to switch off the Office quick loader. RTFM and quit with the FUD already.
As I've mentioned elsewhere in this thread, I also use OpenOffice regularly, on my home PC. I have worked extensively with the word processor and spreadsheet apps, and have given the graphics app a serious try.
And I'm sorry, but I still completely disagree with you about its suitability for businesses. Among the good, and indisputable, reasons to remain with Microsoft Office for now are...
If OpenOffice works for you and does what you need it to, great, I have no problem with your saying that. But please don't go claiming it's ready for everyday business desktop use when it's so easy to point out things like the above. Potential? Yes, it's going to be fantastic one day. Current use? No way, for most office environments. And the latter is what this thread is about.
No, these crimes result in a much more practical impact: they cost health and counselling services a fortune to look after those who get themselves hurt and then need help to recover.
I'm not personally a drug user, but as far as I can see, cannabis is widely used and has far less of a health implication than alcohol or tobacco. There is a very good argument for legalising it, not least that there's no particular reason not to and most people in many countries where it is currently illegal seem to be in favour of the change.
What I'm talking about here is the hard drugs -- the ones that kill, or do serious damage, without a doubt -- which are used by far fewer people, but cause the vast majority of the mess (as well as the links to other organised crime, yada yada).
Prostitution is a whole different thing, but the simple fact is that most of the people involved in that line of work aren't doing it out of choice. They're abused, tricked, blackmailed, feeding drug habits, or otherwise motivated by something other than their personal desire to shag lots for money.
In cases such as those you cite, there are generally specific legal exemptions for those taking part in a law enforcement capacity.
If there were not, everything from high speed police pursuits to arrest using firearms would be illegal. Do you really think that's a good idea?
If only that were so. Alas, it's often the case that people involved in these things, while perhaps (though certainly not always) starting the activity voluntarily, don't continue it out of choice. Moreover, the cost is often to society as a whole, when it has to clear up the resulting mess.
In the UK, it appears to cover the act of copying, full stop. There are specific exemptions and things aren't always black and white, but I'm not aware of anything that requires you to distribute a copy before your actions become illegal. Then again, I'm not a lawyer, I only deal with these issues in relation to some of my hobbies, so please cite anything that proves me wrong; you'd make my life much easier and save my friends and I a lot of money!
If you're going to make claims like that, do please state which "most countries" you're talking about, and whether or not you are a lawyer or otherwise qualified to offer that opinion. What you say is certainly not the case in many western jurisdictions, and with no disrespect intended, your arguments read like a person who's wishing the law worked his way, not an informed legal opinion. I'd love to see you cite test cases in any western jurisdiction and prove me wrong, but I doubt that's possible.
Um... No. What you just wrote is technically infeasible with any of the currently mooted DRM technologies, and there is no legal basis, current or proposed, to support it, or to defend the perps who have "accidents" from accusations of malicious activity. You are spouting panic-induced FUD, and you desperately need to do some homework on what is, and is not, part of the current DRM ideas, both technical and legal.
SourceSafe is a fairly horrible product, it's true, but I guess we'll just have to agree to disagree about Visual Studio. I'm also a professional developer, and have been a "guinea pig" for a move to a new version of Visual Studio twice in recent years.
At first, I didn't like the latest versions. They have their flaws, they certainly could be less bloated, and Visual Studio.NET mysteriously removed a couple of very nice features that were present in VS6. But even ignoring all the .NET hype (which is irrelevant to me since my code runs on 15 different platforms), after keeping an open mind and giving it a chance, I find VS.NET to be a nice development environment. The recently released VS.NET 2003 seems to fix some of the more glaring issue with the original VS.NET, too.
VS.NET is very configurable. It can do all the same things with make files and such as the alternatives you mentioned, and has enough compiler options to make even GCC blush. The editor is pretty decent. The compiler outputs very high quality code compared to the field (and I work in mathematical software, so things like optimisations and generally efficient output code matter here).
I think the top spot for me is the debugger. A lot of my work involves tracking down bugs in complex numerical algorithms, and I make regular use of many of the VS debugger's more advanced features. An AC replied to me suggesting things like Qt Designer. Sure, some people have nicer GUI designers out there, but I'm talking about serious tools, not eye candy generators. What is the best debugger available in the Linux/GCC/LIDE world?
I'd love to know where you got your ideas about what's taught in university these days from, BTW. I'm not in academia myself any more, but several of my friends are in CS research at major UK universities, so I'm pretty well up on the academic scene here. I'm not aware of any group that specifically teaches anything much about the tools you mentioned; most of the universities, for better or worse, seem to use Java as their main teaching language now, and for the scale of project we're talking about there, the tools you mentioned are serious overkill.
Gee, what are those little icons in my system notifications area?
Oh yeah, it's the quickstarters for Mozilla and OpenOffice.
But seriously, we're talking complete packages here. Microsoft do provide an OS and Office combination that work well together. Whether you like their commercial practices or not, businesses can buy into that and gain its advantages today. If Linux/OpenOffice/whatever can't match it, that's their problem, not Microsoft's.
I just love people like you. In a discussion about an article that questions whether there is any reason to buy Microsoft now rather than open source, you're looking at where open source might go later.
The question at hand is whether or not there's still any compelling reason to buy MS software today. The answer, for businesses looking for office software, is yes, for the reasons I and others have given in this thread. We all know open source has a lot of potential, but what happens tomorrow is a different question, and you don't know the answer to that any more than I do.
On the contrary; I find OpenOffice to be quite zippy on my WinXP box, at least comparable to MS Office.
The places OpenOffice is still playing catch-up to MS Office are much more serious at the moment: stability, feature set and ease of use.
I mainly use the word processor, for a variety of different document types from simple letters to DTPing programmes for events to technical reports. I also use the spreadsheet, both for numerical work and as a toy database and mail merge data source.
I have found that both apps are generally less resilient than the MS Office equivalents, particularly in the much-advertised import and export of the de facto standard Microsoft file formats.
There are some features that apparently aren't properly or completed supported, at least not yet, though for most everyday uses either app is adequate for most people.
Perhaps the biggest issue is all the "little things". MS Office has had a lot of usability research done on it. You may not like the paperclip, adapting menus or wiggly spellchecker underlines, but you can switch them off. However, to pick on styles as a usability example, OpenOffice has a nice theoretical mechanism, but where are my keyboard shortcuts for styles and special characters in OpenOffice? Where's the key to reset the character format back to the paragraph style's default? How do I reset a paragraph style to not specify any font changes from the base style? There is functionality approximating most of these things, but it just isn't quite tidy somehow. You can cite similar problems with many other specific areas of the code.
It's got great potential, but one good answer to the question "Any reason to buy Microsoft?" is simply yes, because (for now at least) it's still quite a way ahead on usability.
Quoth the AC:
If you had R'd TFA then you'd spot that a lot of it does relate to things other than the OS side of the market, and many of the claims made are general and across the board. In fact, from the original article, and citing a guy from SuSE of all places:
Some of the other more telling quotes from the article follow.
That is still true, if anything more so today than it was five years ago. MS still totally dominate the desktop. In particular, their Windows development tools and office suite still completely outclass the OS equivalents. To give credit where it's due, a couple more years at this rate and OpenOffice could be a real threat. I haven't seen any open source project currently in development that's even close to Visual Studio.
Because of course, open source things are immune to bugs and security problems... not. If you really think "almost everything" that was once unique to Microsoft now has a serious open source competitor, you haven't been looking very carefully.
It may be a competitor, but it ain't a better product. It's got a way to go before it challenges either the raw power or the ease of use of the Microsoft suite. For geeks who are happy to play with new toys, it's great, and maybe in the future it will be great for Joe Average as well, but enough with the kidding ourselves, OK? It isn't there yet.
OK. I claim that operator overloading can be used sensibly, and that contrary to claims of its detractors, this seems to happen most of the time even in C++. I further claim that when used sensibly, operator overloading significantly increases readability and maintainability. It is also a necessary feature to co-ordinate with writing generic algorithms using tools like C++'s templates. (Yeah, they have nasty syntax at times, but they're way better than nothing.)
Of course you can be sure. You can't overload operator+(char*, int) since it takes two built-in types and overloaded operators must use at least one UDT. You can overload operator&&, which in my view is a flaw in the current C++ scheme, but this is a well-known issue and standard practice is not to do it.
If you really think operator overloading is a bad thing, perhaps we should all give up forming strings using "Hello, " + name + ", how are you?" and go back to good ol' strcat and friends? Of course, Perlesque interpolation ("Hello, {name}, how are you?") is often cleaner still in this particular case, but one thing at a time! :-) (And it has its own disadvantages too.)
In general, the overloading of arithmetic operators for numeric classes such as matrices or complex numbers, the overloading of indexing operators for container classes, the overloading of increment/decrement/dereferencing operators for iterators and such all lead to code that looks more consistent than if you have a mishmash of arbitrarily named functions for each type to do the same thing.
As long as the person who writes the overloaded operators does so sensibly -- and you can write any function to have stupid side-effects, so this is no different to writing any other code -- the benefits in consistency are valuable aids to readability and maintainability. Don't look at what can be in the worst case, look at what can be in the best case, and most important of all, what actually happens. In this case, history amply demonstrates that operator overloading can be of genuine value in real world projects.
I'm in a similar position. I work for a company where I guess the average developer is 25-30, the average team leader 30-35 and there are several significantly older guys around. We have few managers; the older guys still write code like the rest of us for most of their time.
I saw a set of figures once (in the Mythical Man Month or something similar? Can't recall...) for the productivity of a developer. Basically, it claimed that over-30s could successfully work on more than one project at once. Over-25s were OK on one project, but not good at multitasking. Under-25s were pretty much only useful for documentation; give them a serious development role to play and they do more harm than good.
While I don't entirely agree with this from personal experience, I can well believe that taken across the whole industry these figures are a reasonable estimate. Certainly a lot of just-graduated kids think they're clever and dive in trying the latest and greatest tools, and promptly make a huge mess that the older and wiser then have to clear up (while still doing their own stuff as well, of course). Meanwhile the more mature and experienced developers are much more sceptical of new shiny things, and prefer tried and tested tools and techniques until new ideas have been proven in practice. That in itself makes them much less of a liability than the go-faster-stripes generation.
There is a natural problem with many industries, software development included, that the Peter Principle applies. Coders are promoted to their level of incompetence. Often they're also transferred out of development and into management, not because they're going to be more useful there or have shown any particular aptitude for management, but simply because they're older and the existing stupid managers think that doing this will help.
I've noticed in my time in the industry that where you have people who are still close-to-the-metal developers in their late 30s -- by which time many have dropped into the management pool -- and beyond, it's usually because they've made an active decision to remain that way, and they're usually the kind of people you would want to work with as a developer, or to whom you could entrust your vital project as a manager. These people should be given the respect they deserve and treated like the gold dust that they are, not sidelined because their DoB is earlier than 1980. Most of the more successful businesses I've seen, particularly in hard times such as the past couple of years, seem to take older developers seriously. Coincidence? I think not.
But what would I know? I'm only 25, and just about able to work on a single development project without destroying it. ;-)
Why? Although I've never used it "seriously" myself, even I can see its obvious attraction for some types of job, and that it would be more appropriate than C++, Java, Python, or whatever other alternative. As with anything, you should pick the right tool for the job, whether or not it happens to be from Microsoft and whatever its reputation amongst certain types of programmer.