C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  C++ Templates - feelings honestly expressed Goto page 1, 2, 3 ... 11, 12, 13  Next          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Harald Nowak Guest Posted: Wed Apr 07, 2004 7:41 pm    Post subject: C++ Templates - feelings honestly expressed As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. Don't get me wrong on that: templates are senseful! Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. Many former die-hard C++ programmers have already left the C++ train and jumped upon the Java train because of that very reason. What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Any opinions welcome! [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Thomas A. Horsley Guest Posted: Thu Apr 08, 2004 10:48 am    Post subject: Re: C++ Templates - feelings honestly expressed Quote: What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Any opinions welcome! I know nothing about market share, but I do know I've learned how to make template code readable (for me anyway): I happen to think Ada's concept of "instantiating" generics makes a lot of sense, so I do something similar with C++ templates. Whenever I need to use something like a template class, I make a typedef with a sensible name I can use in place of something that looks like <<ba><r>::<>$#@! (C++ template syntax is still more readable than common lisp macros though . Of course that doesn't solve my biggest objection to lack of explicit instantiation: You can accidentally pass the wrong argument to a function, and "shazam!" behind your back some template function you didn't even realize existed gets instantiated, the program compiles with no errors and just does some totally weird and unexpected thing, and it is weeks before you figure out what is going on. Quote: there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. I think the problem here is more the manuals than the templates. Has anyone yet published a good modern C++ book that utterly ignores the existance of all the old variations of C++ libraries and non-standard pre-release template libraries? I keep wanting a good book I can use to make sure I write standard conforming programs, but everywhere I look all the books continue to document all sorts of backward compatible versions of streams libraries, and template classes that aren't in the STL but might be someday, etc. I don't want this stuff in the book - I want a book that no matter where I look, I'll only see standard conforming information . -- Quote: ==>> The *Best* political site <URL:http://www.vote-smart.org/> >>==+ email: [email]Tom.Horsley (AT) worldnet (DOT) att.net[/email] icbm: Delray Beach, FL | <URL:http://home.att.net/~Tom.Horsley> Free Software and Politics <<==+ [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Andreas Huber Guest Posted: Thu Apr 08, 2004 10:50 am    Post subject: Re: C++ Templates - feelings honestly expressed Harald Nowak wrote: Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. You seem to suggest that an average coder should be able to understand all code written in a particular language. Moreover, if that is no longer the case, you seem to suggest, the language is broken in some sense and should be abandoned. I've heard that logic for several times now and I really don't see why programmers think this way. For non-templated libraries in any language, programmers readily accept that they often cannot even get hold of the implementation source code. Even if they can, I very much doubt that the average programmer would be able to readily understand complex libraries like e.g. for regular expressions. Sure, everyone *can* read such code and, on a very low level, explain what's happening (add x to this variable, take this substring and add it to that string, etc.). However, it's an altogether different matter to explain what's happening on a much higher level. Code has been like this for decades and programmers have been accepting that they cannot understand everything. Note that it's exactly the same situation for template metaprogramming code, many programmers can explain what's happening on a very low level (specialize this template for the case when the type is equal to that type, etc.) but most fail to explain what a complex metaprogramming construct is good for on a much higher level. The only thing what an average programmer should be able to grok is how to *use* a templated *interface* but I fail to see why people think they must also be able understand all the metaprogramming wizardry hidden behind that interface. Quote: Don't get me wrong on that: templates are senseful! Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. I'd say that the library is probably badly designed and/or documented. Please provide an example. [snip] Regards, Andreas [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top David Abrahams Guest Posted: Thu Apr 08, 2004 10:52 am    Post subject: Re: C++ Templates - feelings honestly expressed [email]Harald.Nowak (AT) web (DOT) de[/email] (Harald Nowak) writes: Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. Don't get me wrong on that: templates are senseful! Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. Many former die-hard C++ programmers have already left the C++ train and jumped upon the Java train because of that very reason. What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Any opinions welcome! Templates can be used for good or evil. With power comes responsibility. Some of the heaviest template code in libraries is used to make efficient programs far _more_ readable. See for example the two Finite State Machine examples attached to http://article.gmane.org/gmane.comp.lib.boost.devel/35273, and consider everything preceding the declaration of class player to be part of the FSM library framework. The language used for describing the player FSM to the library is almost a direct transcription of the state transition table you might write when building it by hand. There are numerous other examples of such domain-specific embedded languages, c.f. Boost.Lambda, which allows: // print each element of a multiplied by pi std::for_each(a.begin(), a.end(), std::cout << (_1 * 3.14159)); or Boost.Spirit ([url]http://spirit.sf.net)[/url], which allows one to easily write parsers directly in C++ code. -- Dave Abrahams Boost Consulting www.boost-consulting.com [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Marco Oman Guest Posted: Thu Apr 08, 2004 6:02 pm    Post subject: Re: C++ Templates - feelings honestly expressed Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. Don't get me wrong on that: templates are senseful! I agree on the first part: things like Boost' MPL are very hard to read and understand. But I do not see here a reason to change the language I use to solve my problems. After all templates obeys the rule "you do not pay for what you not use", and so I can go on doing my work with the rest of the stuff as usual. What may hurt me is that now I get a growing feel of being unable to master a section of the language (or perhaps is more correct to say a new family of patterns supported by trhe language). This has (among the others) two implications: one is simply self estimation, the other is "quotation" on the job market. And could be reduced to: shall I change language because it is easier for me to be a 100% Java programmer rather than a 100% C++ one? Okay, the question is put rather hard, the mood culd rise flame wars, anyway I think hard-code template programming is an area worth to be explored. Even if I can't (because I do not have the time or the functional programming background) to do it. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Uwe Schnitker Guest Posted: Thu Apr 08, 2004 6:11 pm    Post subject: Re: C++ Templates - feelings honestly expressed [email]Harald.Nowak (AT) web (DOT) de[/email] (Harald Nowak) wrote in message news:<ceb032e9.0404070730.4130a66e (AT) posting (DOT) google.com>... Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. Well, Stroustrup used to answer such claims with "When I introduced classes into C++ (rsp. C with Classes), people complained about how useless, complicated and difficult to use they were". I think he's right on that, and something like "nearly impossible to read and understand for the average coder" might have been written then, too. Ok, it is a bit worse wrt. templates, since they are a bit more difficult to read - not because they were designed badly, and not just because of overuse, but because the syntax design space for them was more limited. Quote: Don't get me wrong on that: templates are senseful! Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. Well, many seasoned C programmers had the same problem when switching to classic C++. People who learned C++ in the last few years - and were not spoiled by to much Java exposure before - have considerably fewer such problems. Modern C++ is a different, and IMHO far better, language than classic C++. Quote: Many former die-hard C++ programmers have already left the C++ train and jumped upon the Java train because of that very reason. Many former C++ -> Java switchers are coming back to C++ because it has evolved into such a wonderful language. Quote: What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Any opinions welcome! IMHO, the biggest problem is the lacklustre support for modern C++ programming styles offered by todays tools. Debugging tools get better, but leave much to be desired. "Class Browsers", as their name already implies, don't get along well with free functions - a must with templates - and code completion is weak with them, and refactoring tools which are template-savvy - e.g. allowing almost-common code to be refactored into template functions - are not even on the radar screen yet. Today, tools have far more problems with templates code than programmers. Don't get me wrong, todays tools can process template code quite well, but they fail to present it to programmers in a really enabling way. Since for todays projects the scope often is very large, so that tool support for programmer insight is crucial, this is the real bottleneck. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top kanze@gabi-soft.fr Guest Posted: Thu Apr 08, 2004 6:16 pm    Post subject: Re: C++ Templates - feelings honestly expressed [email]Harald.Nowak (AT) web (DOT) de[/email] (Harald Nowak) wrote in message news:<ceb032e9.0404070730.4130a66e (AT) posting (DOT) google.com>... Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. It's the current "in" thing. When I first started C++, OO was the current "in" thing, and the usage of virtual functions and derivation went totally overboard. Both templates and virtual functions are useful tools. I'd hate to be without them. That doesn't mean that the are the solution to everything, or that they should be used everywhere. People being what they are, however, whatever happens to be "in" will be abused and overused. It's not a C++ problem, and there's not much C++ can do about it. Quote: Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. It's certain that experimental coding techniques will not be understood by those unfamiliar with them. And that before introducing them in a project, you really do need to do a costs-benefits analysis -- retraining programmers to be familiar with them may be a lot of cost for relatively little gain. Or a big gain, if you happen to need the new techniques -- it all depends. Long term, you have to program some sort of evolution, or you'll be left behind. OO had similar problems 10 or 15 years ago, and any number of programmers refused to consider it because it obfuscated code. (When you write p->f(), who knows what function actually gets called. It's worse than generic programming, really, because the answer can't be determined by any amount of static analysis.) IMHO, we are right at the beginning of the learning curve with regards to generic programming. Something like where we were 25 or 30 years ago with OO. Which means that intensive generic programming probably isn't appropriate for most mainline applications. Yet. (Which doesn't mean that you can't use third party libraries which use extensive generic progrramming in their implementation.) Quote: Don't get me wrong on that: templates are senseful! Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. That sounds like a problem with the library documentation. I, and my collegues, don't do much generic programming ourselves, but we have no trouble using std::vector. Quote: Many former die-hard C++ programmers have already left the C++ train and jumped upon the Java train because of that very reason. And many former C programmers still stick with C. New technologies introduce new challenges. One of the real advantages of C++ is its backwards compatibility -- you can move into the new technologies gradually. If you don't want to abord intensive generic programming just yet, you don't have to. But you can play around with it in some non-critical parts of the code without changing languages. (And I have a great deal of problems understanding the move to Java. I gave Java a try, and have pretty much left it because it doesn't support any real OO.) Quote: What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Any opinions welcome! I often think that C++ is moving to quickly. I think that there is room for discussion concerning the speed. But move it must, or it will die. Personally, I think that after a short hiatus due to the Java mode, C++ is back in the driver's seat. It isn't gaining users at the rate it initially did, simply because there aren't that many users left to gain; I can't imagine a professional programmer today not being competent in C++. Not necessarily mastering all of the subtilities, perhaps, but capable of writing robust, working programs without too much problem. IMHO, most of the problems with C++ are due to the past, and not the future. If I don't like meta-programming, I can use C++ without it. If I find the rules of operator overloading too complex (I do), then I just avoid overloading functions which have different semantics, so the results of overload resolution cannot affect the correctness of the program. On the other hand, don't get me started about the preprocessor or the declaration syntax. -- James Kanze GABI Software mailto:kanze (AT) gabi-soft (DOT) fr Conseils en informatique orientée objet/ http://www.gabi-soft.fr Beratung in objektorientierter Datenverarbeitung 9 place Sémard, 78210 St.-Cyr-l'École, France, +33 (0)1 30 23 00 34 [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ben Hutchings Guest Posted: Fri Apr 09, 2004 12:12 am    Post subject: Re: C++ Templates - feelings honestly expressed Thomas A. Horsley wrote: <snip> Quote: I think the problem here is more the manuals than the templates. Has anyone yet published a good modern C++ book that utterly ignores the existance of all the old variations of C++ libraries and non-standard pre-release template libraries? snip I think that's exactly what Josuttis's book does: <http://www.josuttis.com/libbook/>. I haven't read it but it is highly recommended by others. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Walt Karas Guest Posted: Fri Apr 09, 2004 12:31 am    Post subject: Re: C++ Templates - feelings honestly expressed The vast majority of the hard-to-follow code I've seen was the result of apathy or poor impulse control: the author just didn't want to spend the time thinking about strategies/approaches to make the code more compact, organized and comprehensible. I've worked in several progamming languages, but have yet to see one that prevents the writing of hard-to-follow code. When code is hard to follow because the author uses different strategies/approaches than those you are familiar with, at least you get the mental stimulation of learing something new, whether you adopt it or not. My reaction to most extreme preprocessor/template metaprogramming is that I'd rather use a more powerful macro processor (like m4), or a script or C++ program that generates C++ code. But alot of people feel macros are inherently evil, and that code that generates code is even more weird and objectionable than metaprogramming. They'd rather just open their editor and dive into the multi-hour cut-and- paste frenzy. Ultimately, you just have to learn to tolerate it that what's being done in the next cubicle may be different that what's being done in yours. Any cure I know of for this would be far, far worse than the disease. When using the STL, I prefer to refer to C++PL or some other tutorial. I only look at the standard as a last resort. You seem to feel this is very undesirable, but I don't see why. I also do not find the way the Standard specifies the contraints on STL type parameters to be very intuitive. In a template I wrote for the public domain ( http://www.geocities.com/wkaras/gen_cpp/avl_tree.html ), I used "references classes" to specify the contraints, which doesn't force the reader to learn alot of new notation. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Nicholas Salerno Guest Posted: Fri Apr 09, 2004 1:01 am    Post subject: Re: C++ Templates - feelings honestly expressed [email]Harald.Nowak (AT) web (DOT) de[/email] (Harald Nowak) wrote in message news:<ceb032e9.0404070730.4130a66e (AT) posting (DOT) google.com>... Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. Don't get me wrong on that: templates are senseful! Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. Many former die-hard C++ programmers have already left the C++ train and jumped upon the Java train because of that very reason. What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Any opinions welcome! I have similiar feelings too. I like the metaprogramming (or parametric programming to other) paradigm. It is interesting. I thought that it would be great for numerical processing applications. I did not want to be academic about it, I actually tried to use it at work on real projects. For image procesing algorithms it showed some promise. I started using it for some computer vision applications. However, as you have stated, others do not want to invest into something they did't know. Also, they did not want code that they would have to maintain in the event I am no longer employed. It didn't matter that the code worked and was efficient, they just didn't know C++ templates. We are talking about C and/or C++ via MFC programmers so I understood their concern. However, as soon as I started to take C++ metaprogramming beyond scientific applications and into general purpose applications I, just like you, found heavy use of templates complicating the project. This is not because of templates in general. I am sure that a clean template language would handle the complexity of templates with good syntax and grammar. The complications arouse from C++'s implementation of templates. For example, I read the example on using type lists to add visitor pattern behavior to an existing object hierarchy. While I was able to understand after reading the article three times, I can definitely see myself not knowing what that code is doing six months from now. Of course one forgets all code not worked on in six months time. What I am saying is that it would take longer than normal to refresh myself. Now I understood my co-workers fear of C++ templates at a deeper level. Lastly, the more I did general application stuff, especially GUI development, the more I was moving away from templates. I just found them to be limiting in certain situations. For example, take Microsoft's WTL library (not officially supported, something they were experimenting with). The WTL library employs some modern template idioms. The problem I had was that with templates everything is set into stone at compile time. Also, types were defined based on other types at compile time. At runtime, how do I have a list of widgets? Say a parent has a list of all the child widgets? vector<CWidget*>? Perhaps, but there is no such thing as a CWidget, but there is CWidget<typename T>, thus an example would be "class CMyButton : public CWindowImpl<CMyButton>" (iirc) to establish compile-time polymorphism. In conclusion, I have found templates to be great for some domains and bad for other domains. I know that the template brigade will disagree with me on that but that is my feeling and perhaps the feeling of many others as well. Nicholas [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Roland Pibinger Guest Posted: Fri Apr 09, 2004 1:06 am    Post subject: Re: C++ Templates - feelings honestly expressed On 7 Apr 2004 15:41:27 -0400, [email]Harald.Nowak (AT) web (DOT) de[/email] (Harald Nowak) wrote: Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. Many programmers I talked to pointed out to me, that they are not willing to invest any further in that language, since the template mechanisms, especially the more and more widespread meta programming techniques, do obfuscate the code in a way that makes it nearly impossible to read and understand for the average coder. Don't get me wrong on that: templates are senseful! Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. Many former die-hard C++ programmers have already left the C++ train and jumped upon the Java train because of that very reason. What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Any opinions welcome! In general I agree with you. It is amazing that some people put so much effort into this area. But you should distinguish between real-world and experimental (b)leading-edge programming. One month ago Scott Meyers wrote in this newsgroup: "As a general rule, Boost has very little visibility, the Standard not much more (many programmers make no distinction between the behavior of their compiler and the behavior dictated by the Standard), and the STL (which everybody has *heard* of) is still viewed as largely foreign territory". So, template "obfuscations" are not recognized by the average C++ programmer. Boost e.g. essentially produces libraries from specialists for specialists. Mainstream programmers will not accept this amount of complexity. Best regards, Roland Pibinger [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Hyman Rosen Guest Posted: Fri Apr 09, 2004 1:21 am    Post subject: Re: C++ Templates - feelings honestly expressed Marco Oman wrote: Quote: shall I change language because it is easier for me to be a 100% Java programmer rather than a 100% C++ one? You won't be. If you had the capacity to be a 100% Java programmer then you would have the capacity to be a 100% C++ programmer, and vice versa. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Le Chaud Lapin Guest Posted: Fri Apr 09, 2004 1:30 am    Post subject: Re: C++ Templates - feelings honestly expressed [email]Harald.Nowak (AT) web (DOT) de[/email] (Harald Nowak) wrote in message news:<ceb032e9.0404070730.4130a66e (AT) posting (DOT) google.com>... Quote: As a long term C++ user (and aficionado) the feeling starts to creep upon me, that the usage of templates in C++ seems to go totally overboard. [snippage] I agree. Mind you, I am not one to insist that everyone use handsaws because some lumberjack accidentally cut his hand of with a chainsaw, but I do see that there is great imbalance in writing template code. I recently made a post in this group asking for a Monarchy<> Polyarchy<> classes. Apparently, they do not exist in the STL. That's odd, considering how fundamental tree structures are in programming. Far more important than exercising one's wizardry to make the most masssive, all-encompassing, unintuitive templates possible; is first finding balance in the *basic* set of template classes generally offered. But templates are not the only constructs that suffer from this problem. These days, virtual functions too are abused to a sickening degree, and often, the coder has no real reason other than lack of forethought to make all 23 functions of a class virtual. S/he is merely exercising his/her right to participate in the "Hey-Dad-Watch-Me-Use-Virtual-Functions" club. I think this behavior might be underlying some of the template abuse. *Some* complexity is inevitable in templates, but I think a lot of it is people just trying to express themselves through code. -Chaud Lapin- [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Stefan Slapeta Guest Posted: Fri Apr 09, 2004 11:28 am    Post subject: Re: C++ Templates - feelings honestly expressed Harald Nowak wrote: Quote: Even Java has finally found its way to a sensible (and senseful) use of templates to ensure typesafe programming. What in particular makes you think the concept of templates in Java is better than in C++? Quote: But, as I said: even I, a seasoned C++ programmer, start having trouble to decipher and understand C++ template code - there are times when I am reading C++ Library docs and have to read them over and over again, until I get a slight grasp of how I have to use them, ALL because of the usage of template mechanisms they make. I think there are two reasons for your observance: First of all, the documentation, which is far away from being perfect in many libraries because they have been written in quite a short time where documentation didn't have the priority it should. Secondly: Generic programming is no easy concept and templates are just the tool for it! I would say, it usually takes months and years to think "generically", and I'm sure it takes longer than getting familiar with OO. But it's always fascinating! (...except you try to learn all about templates within one day.) Quote: Many former die-hard C++ programmers have already left the C++ train and jumped upon the Java train because of that very reason. Only a fool can change from C++ to Java because of templates. If you have a look at industrial trends, the opposite is the case: many who had proudly jumped on the Java train now come back to 'conventional' languages for several reasons. Quote: What do you say - isn't there a real danger of loosing steam (and market shares) for C++ because of that template (mis)use?! Why so? Who is forced to use templates if he doesn't want to? <cynism> There are still many people who write C++ code that differs from C just in the ending of the source file, I don't think many of them are scared because they don't know anything about templates, either :-O </cynism> S. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Stefan Slapeta Guest Posted: Fri Apr 09, 2004 11:38 am    Post subject: Re: C++ Templates - feelings honestly expressed Thomas A. Horsley wrote: Quote: I don't want this stuff in the book - I want a book that no matter where I look, I'll only see standard conforming information . It seems what you are looking for is a book named "ISO/IEC 14882:2003 Programming Language C++", also known as "The C++ Standard". S. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First         C++Talk.NET Forum Index -> C++ Language (Moderated) All times are GMT Goto page 1, 2, 3 ... 11, 12, 13  Next Page 1 of 13   Jump to: Select a forum C/C++----------------C LanguageC Language (Moderated)C++ Language (Moderated)C++ language (comp.lang.c++)C++ language, library and standards Non-English C/C++ Newsgroups----------------C++ (German)C++ (French)  You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum Powered by phpBB © 2001, 2006 phpBB Group SEO toolkit © 2004-2006 webmedic.