C++Talk.NET C++ language newsgroups   Archives   FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in  Library/component design and implementation in C++ Goto page 1, 2, 3, 4, 5, 6, 7, 8  Next          C++Talk.NET Forum Index -> C++ Language (Moderated) View previous topic :: View next topic   Author Message Kymert persson Guest Posted: Wed Aug 27, 2003 9:40 am    Post subject: Library/component design and implementation in C++ Hi, I was wondering if there are any more C++ books along the lines of "Large scale C++ software design" by Lakos, J That is, a book focusing on library/component design and implementation in C++ ("Programming in the large") without a strong bias toward C++ specific techniques such as templates (which I know is dealt with in several other books). I like this book because i feel it directly adresses the the _really difficult_ problem I face everyday as a C++ programmer, i.e. to design and implement libraries/components having the desired properties in terms of functionality, efficiency, maintainability etc. Since I read this book I have performed quite an extensive literature survey for books having this particular focus. To my great surprise I have found no books that really satisfies this criteria, instead I find a lot of books concerning one or more of the following topics: 1. The C++ language itself. 2. "Programming in the small" design issues in C++ E.g. the effective series by Meyers, S. 3. (C++) Patterns and idioms. (E.g. "Design patterns - elements of reusable software". This is slightly off target since these books presents good design rather than how to achieve such designs.) 4. C++ component design and implementation using templates and the STL. (These are also slighlty off target due to the strong bias toward templates, whereas there is still a lot of component design issues that are dificult and does not involve templates, e.g. interface design) 5. General literature on software engineering and design. So far I have found no book in this category that has really helped me to design better libraries/components, typically I find them to be much too abstract to really help at a technical level, i.e. when you start to think about actual classes, interfaces etc. (And don't get me wrong here, there are a lot of great books in the above categories as well, but i think they lack the particular focus I am searching for.) Does anyone know of such literature? Any thoughts, suggestions and/or comments on this issue are greatly appreciated! Thanks in advance. /Persson, K. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top smitty_one_each Guest Posted: Thu Aug 28, 2003 2:30 pm    Post subject: Re: Library/component design and implementation in C++ I'd like to append to the excellent parent post a request for "reference projects" (if you will) on sourceforge.net. Projects that are poster children for good design, coding style, and documentation. Projects that use boost libraries especially. The Linux Development Platform http://www.faqs.org/docs/ldev/ is a good book, with a lot of information in a small package. Applied C++: Practical Techniques for Building Better Software http://www.awprofessional.com/isapi/product_id~{B4EA13D6-62B7-440B-8495-6D4AC4A7E105}/catalog/product.asp is great, but neither book (nor any others I've read) discuss much about laying out a project. I guess such a discussion exceeds the bounds of the language, and quickly gets mired in platform detail. Oops. Thus, looking at existing projects seems like a good approach to finding out which dogs really hunt. [ 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 Aug 28, 2003 7:54 pm    Post subject: Re: Library/component design and implementation in C++ "Kymert persson" <mikaelt (AT) ida (DOT) his.se> wrote Quote: Hi, I was wondering if there are any more C++ books along the lines of "Large scale C++ software design" by Lakos, J That is, a book focusing on library/component design and implementation in C++ ("Programming in the large") without a strong bias toward C++ specific techniques such as templates (which I know is dealt with in several other books). Ok, so "Modern C++ Design" is not what you are looking for. (But I hope you read it anyway, or at the very least read the chapter - available online - about "Policy Based Class Design". It's important to understand that Andrei didn't start off one day "Let's become Mr. Template", but he did end off (Some native speaker care to comment on that, out there?) with it when trying to solve some very concret problems.) Quote: I like this book because i feel it directly adresses the the _really difficult_ problem I face everyday as a C++ programmer, i.e. to design and implement libraries/components having the desired properties in terms of functionality, efficiency, maintainability etc. One book that comes to mind is Carrol/Ellis "Designing and Coding Reusable C++". It should fit your bill exactly, but it's quite dated - written the same time as Lakos' book, I believe. If you keep in mind that you have to adapt your findings to use more modern C++ styles and techniques - the same way you have to do with Lakos' book - it may be very useful. Quote: Since I read this book I have performed quite an extensive literature survey for books having this particular focus. To my great surprise I have found no books that really satisfies this criteria, instead I find a lot of books concerning one or more of the following topics: 1. The C++ language itself. 2. "Programming in the small" design issues in C++ E.g. the effective series by Meyers, S. 3. (C++) Patterns and idioms. (E.g. "Design patterns - elements of reusable software". This is slightly off target since these books presents good design rather than how to achieve such designs.) 4. C++ component design and implementation using templates and the STL. (These are also slighlty off target due to the strong bias toward templates, whereas there is still a lot of component design issues that are dificult and does not involve templates, e.g. interface design) Some books which describe the usage as well as the design of existing C++ libraries might be useful for your purpose, too. E.g. the two book about "C++ Network Programming" from Schmidt/Huston. IIRC, especially the second one describes how to use patterns, frameworks and components to create - in this case - a distributed computing library. Another one would be "The Boost Graph Library", but it describes a generic library, so you might object to "template-heavyness". But generic components/ libraries are a very important concept in modern C++. Well, you can't really avoid what you called in your opening paragraph "a strong bias toward C++ specific techniques" if you want to use C++ effectively. A "large scale C++ solution" will typically look quite different from a "large-scale (Java|Perl|Phyton|Ruby) solution", even if the fundamental problems - looked upon from a high level - are similiar. Quote: 5. General literature on software engineering and design. So far I have found no book in this category that has really helped me to design better libraries/components, typically I find them to be much too abstract to really help at a technical level, i.e. when you start to think about actual classes, interfaces etc. (And don't get me wrong here, there are a lot of great books in the above categories as well, but i think they lack the particular focus I am searching for.) One book that doesn't have this problem is Robert Martins "Agile Software Development". It has lot's of examples in C++ and Java. It is also a book about patterns, but unlike those in your category 3 it describes "how to arrive there". While a lot of the content is about "small" issues, there's also a strong focus on large-scale issues. One part of the book deals explicitly with developing a framework - e.g. that one should develop a framework concurrently with several applications, and not beforehand. (You may want to read my review of this book on the ACCU website. Please note that I wanted to mark it as "Highly Recommended", but the ACCU people refused to do so. I also think the categorization "management" is quite stupid.) Quote: Does anyone know of such literature? Any thoughts, suggestions and/or comments on this issue are greatly appreciated! Thanks in advance. /Persson, K. HTH, Uwe [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Kymert persson Guest Posted: Fri Aug 29, 2003 11:01 am    Post subject: Re: Library/component design and implementation in C++ "smitty_one_each" <smitty_one_each (AT) hotmail (DOT) com> wrote Quote: I'd like to append to the excellent parent post a request for "reference projects" (if you will) on sourceforge.net. Projects that are poster children for good design, coding style, and documentation. Projects that use boost libraries especially. I totally agree with you. It would be even greater if the open source guys/gals would document the core design issues within their respective projects so that anyone could benefit from these examples without diving into the source code itself. Quote: The Linux Development Platform http://www.faqs.org/docs/ldev/ is a good book, with a lot of information in a small package. Well this seems to be a very interesting book on establishing a software construction environment in a Unix/Linux (or Unix-like) environment. Thanks for the tip. But at first glance this reference does not seem to be concerned with large scale software and library/component design. Quote: Applied C++: Practical Techniques for Building Better Software http://www.awprofessional.com/isapi/product_id~{B4EA13D6-62B7-440B-8495-6D4AC4A7E105}/catalog/product.asp is great, but neither book (nor any others I've read) discuss much about laying out a project. This one was new to me and it seems to along the lines of what I am looking for. Thanks! Quote: I guess such a discussion exceeds the bounds of the language, and quickly gets mired in platform detail. Oops. Hmm. I do not quite agree with you here. In the general case it is wise to design for potability, so that the main body of your application can be ported to other/new platforms without rewriting the whole system. Often this is done by having lower level platform dependent componets which in turn supply platform independent interfaces to higher levels. So apart from these lower level components you are, ideally, not very dependent on this or that particular platform. Anyway, thanks again for the tips above! /Persson, K. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Kymert persson Guest Posted: Fri Aug 29, 2003 11:04 am    Post subject: Re: Library/component design and implementation in C++ Thanks for several great tips! A few comments though. a) _"Modern C++ Design"_ I have this one as well and think it's an excellent book concerning specific C++ techniques (in this case mainly patterns and their imp. in C++). But again, i think this is a book mainly on "programming in the small" as opposed to "programming in the large". b) _"Designing and Coding Reusable C++"_ Excellent book as well, but suffers from the same "problems" as a) above i think. c) _"C++ Network Programming", "The Boost Graph Library", and "Agile Software Development"_ All these references seem very interesting! Thanks a lot! /Persson, K. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Francis Glassborow Guest Posted: Fri Aug 29, 2003 3:48 pm    Post subject: Re: Library/component design and implementation in C++ In article <Llr3b.26057$dP1.64461 (AT) newsc (DOT) telia.net>, Kymert persson <mikaelt (AT) ida (DOT) his.se> writes Quote: OK then, let me rephrase my original question as follows: Why is it difficult to find __concrete__ C++/software engineering (SE) literature concerning _large scale_ design issues and library/component design? That is, the stuff C++ programmers have to deal with everyday in the "school of experience". Possibly because those who understand it are fully employed and have no time to write books. BTW writing a book is NOT a way to become rich, rather the contrary. I can remember Bjarne Stroustrup telling me that the C++PL 3ed would just about pay for one of his children's university education. OK so Universities are not cheap to attend but the book is one that very substantially outsells almost anything else in the programming field. -- Francis Glassborow ACCU 64 Southfield Rd Oxford OX4 1PA +44(0)1865 246490 All opinions are mine and do not represent those of any organisation [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Kymert persson Guest Posted: Sat Aug 30, 2003 11:24 am    Post subject: Re: Library/component design and implementation in C++ Yes. This is probably a quite accurate description of the state of things in the C++/SE community. I remember this talk i went to by some guy from industry who just had started on his doctoral thseis on open source software. He said that when he started reading the academic literature, he did not recognize himself at all. Really sad if this is the case. I have to admit i haven't checked out the core research literature out there (journal/conference articles etc.), but my guess is that most of it is pretty far out, at least from a practical C++ point of view. /Persson, K. "Francis Glassborow" <francis (AT) robinton (DOT) demon.co.uk> wrote Quote: In article <Llr3b.26057$dP1.64461 (AT) newsc (DOT) telia.net>, Kymert persson [email]mikaelt (AT) ida (DOT) his.se[/email]> writes OK then, let me rephrase my original question as follows: Why is it difficult to find __concrete__ C++/software engineering (SE) literature concerning _large scale_ design issues and library/component design? That is, the stuff C++ programmers have to deal with everyday in the "school of experience". Possibly because those who understand it are fully employed and have no time to write books. BTW writing a book is NOT a way to become rich, rather the contrary. I can remember Bjarne Stroustrup telling me that the C++PL 3ed would just about pay for one of his children's university education. OK so Universities are not cheap to attend but the book is one that very substantially outsells almost anything else in the programming field. [ 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: Tue Sep 02, 2003 6:01 am    Post subject: Re: Library/component design and implementation in C++ In article <3f50b0e1.1453525 (AT) news (DOT) utanet.at>, Roland Pibinger wrote: <snip> Quote: The following links may contain some useful hints allthough they are more style guides than component design books: http://hem.fyristorg.com/erny/industrial/ http://pcroot.cern.ch/TaligentDocs/TaligentOnline/DocumentRoot/1.0/Docs/books/WM/WM_1.html http://www.symbian.com/developer/techlib/papers/cppapi/cppapi.htm Since you are interested in "Programming in the large" take a look at the Java platform especially J2EE. The architectures and design patterns discussed there are mostly language independent. Symbian's conventions are pretty weird because they don't use exceptions [1] or the standard library [2] and they go through some contortions to save memory (particularly stack memory). The Symbian paper is therefore a mixture of basic C++ design guidelines and SymbianOS-specific information, and much of it is not generally applicable. However, the binary compatibility information from Taligent and Symbian might be very useful for those who are concerned about the ability to update one of a set of linked modules in an installed system. BC is a tough problem that's out of the scope of the standard language. [1] IIRC, they decided that general-purpose exception-handling would increase code size too much and that they could do a better job with setjmp-based cleanup of things with non-trivial destructors. Functions that can throw are distinguished by a name suffix rather than by a throw specification. [2] I think EPOC32 predates much of the standard library; besides which, it would clash with their memory management and error-handling style. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Glen Low Guest Posted: Wed Sep 03, 2003 7:30 am    Post subject: Re: Library/component design and implementation in C++ Quote: Isn't it odd that modern C++ is so unhelpful for everyday programmers? That too has puzzled me. When I first saw STL, I said "of course!" -- this must be the next jump in productivity, after objects and components. I'm sure the guys who eventually enshrined it in the C++ Standard Library thought so too. A higher order of programming, much like SQL, with great expressiveness and without a loss of performance. Yet Microsoft ATL is still stuck in macro-land. Java and C# proliferate without the use of generics. Entry level computer courses never talk about templates. The little Mozilla code I've seen is just COM-based interface design. Most of my fellow coders are either comfortable with their large object hierarchies, or struggle to understand how to map real world entities to objects, or are forced to use code generators and/or lots of hard-to-maintain hand-written boilerplate code. I for one, would love to see a proper GUI framework designed using modern C++ generic principles -- if I have the time, I would probably work on it. For example, there are many GUI objects that contain other objects e.g. list views, tree views -- and it is a royal pain to get them to work in a generic fashion, from Java WFC to C# WinForms. Most generic code I know of is non-interface work, which isn't as "glamourous" I suppose. Perhaps C++ template syntax hampers it -- complicated template use throws even doxygen off and different compilers, so if machines have trouble understanding it, what more other humans? Perhaps it is an education issue, how do you explain templates and generic programming in a single sentence? Perhaps we need the killer app, which I think would be a GUI framework. Are Java and C# and Objective-C playing catch-up, or has the time for C++ passed? Cheers, Glen Low, Pixelglow Software www.pixelglow.com [ 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: Wed Sep 03, 2003 9:47 pm    Post subject: Re: Library/component design and implementation in C++ In article <9215d7ac.0309022037.59bf13a5 (AT) posting (DOT) google.com>, Glen Low wrote: Quote: Isn't it odd that modern C++ is so unhelpful for everyday programmers? That too has puzzled me. When I first saw STL, I said "of course!" -- this must be the next jump in productivity, after objects and components. I'm sure the guys who eventually enshrined it in the C++ Standard Library thought so too. A higher order of programming, much like SQL, with great expressiveness and without a loss of performance. Yet Microsoft ATL is still stuck in macro-land. It also makes extensive use of templates, though quite a lot of that is use of the curiously-recurring template pattern. I found it a pleasure to use compared to MFC's COM support; also, much less bloated. But then, that's not saying much. Quote: Java and C# proliferate without the use of generics. Yes, though there has been demand for them; they will be in Java 1.5 and Microsoft Research has added them to a research version of C#. <http://developer.java.sun.com/developer/technicalArticles/releases/generics/> <http://research.microsoft.com/projects/clrgen/> Quote: Entry level computer courses never talk about templates. I can't comment on that; I think I only used C++ once in university and I think that was only prescribed instead of C so that we could use iostreams instead of stdio. Quote: The little Mozilla code I've seen is just COM-based interface design. snip Mozilla's use of C++ has been deliberately conservative because it was expected that it would have to be built with some quite poor compilers. Of course the world has moved on, compilers have improved, and that conversatism isn't so necessary in a project starting now, but it made sense then. [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Ed Avis Guest Posted: Thu Sep 04, 2003 1:36 am    Post subject: Re: Library/component design and implementation in C++ gtkmm <http://www.gtkmm.org/> claims to be a GUI toolkit that works in a modern C++ style, using polymorphism, the STL, namespaces and so on. -- Ed Avis <ed (AT) membled (DOT) com> [ 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: Thu Sep 04, 2003 8:17 am    Post subject: Re: Library/component design and implementation in C++ On 3 Sep 2003 03:30:51 -0400, [email]glenlow (AT) pixelglow (DOT) com[/email] (Glen Low) wrote: Quote: Isn't it odd that modern C++ is so unhelpful for everyday programmers? That too has puzzled me. When I first saw STL, I said "of course!" -- this must be the next jump in productivity, after objects and components. I'm sure the guys who eventually enshrined it in the C++ Standard Library thought so too. A higher order of programming, much like SQL, with great expressiveness and without a loss of performance. What remained from your enthusiasm after you had used STL in a real world project? Probably not much. [...] Quote: Perhaps C++ template syntax hampers it -- complicated template use throws even doxygen off and different compilers, so if machines have trouble understanding it, what more other humans? Generic programming in C++ is way too complicated for users (syntactically and in handling compilation errors). Moreover generic programming has some deficits in principle, eg. it violates encapsulation and lacks extensibility (but, of course, provides configurability). Quote: Perhaps it is an education issue, how do you explain templates and generic programming in a single sentence? If customers don`t buy your product you better change the product and you don`t try to change the customers. Quote: Perhaps we need the killer app, which I think would be a GUI framework. I thought GUI frameworks were a classic case for OO programming? Quote: Are Java and C# and Objective-C playing catch-up, or has the time for C++ passed? Probably the latter. 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 John Potter Guest Posted: Thu Sep 04, 2003 8:21 am    Post subject: Re: Library/component design and implementation in C++ On 3 Sep 2003 17:47:41 -0400, Ben Hutchings <do-not-spam-benh (AT) bwsint (DOT) com> wrote: Quote: Entry level computer courses never talk about templates. I can't comment on that; There is no reason to not talk about templates in the first course. Programming is all about generalizations. Swap is a template that can be introduced in the first course. Simple sorts and searches should be templates. Quote: I think I only used C++ once in university and I think that was only prescribed instead of C so that we could use iostreams instead of stdio. That is the first step in removing the most common error of using C and needing to talk about address-of to do simple input. If it was an advanced course rather than a first course, that is a shame. John [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top John Potter Guest Posted: Thu Sep 04, 2003 8:21 am    Post subject: Re: Library/component design and implementation in C++ On 3 Sep 2003 17:50:53 -0400, llewelly <llewelly.at (AT) xmission (DOT) dot.com> wrote: Quote: glenlow (AT) pixelglow (DOT) com (Glen Low) writes: Isn't it odd that modern C++ is so unhelpful for everyday programmers? That too has puzzled me. When I first saw STL, I said "of course!" -- this must be the next jump in productivity, after objects and components. Many people said that about Object Oriented Programming in the middle 70's . That was quite a few years after Simula-67 first emerged, but about 15 years before OOP got picked up as 'the next big thing'. Of course, generic programming goes back way before C++ templates, but the point is that powerful ideas can take a long time to spread. There's an old saying (origin unknown to me): 'Don't worry about people stealing your ideas. If your ideas are any good, you'll have to ram them down other people's throats!' There's also a curious inversion - *good* generic programming is something one uses far more often than one writes - by definition. I don't think that's true of OO. Interesting. STL is a bit of generic programming that is used often. OO as ADT is used in STL, but inheritenance is almost absent. The problem may be trying to bend STL algorithms and containers to match a problem in the same way that OO tries to derive a new solution from an old class. I doubt that either really works. John [ See http://www.gotw.ca/resources/clcm.htm for info about ] [ comp.lang.c++.moderated. First time posters: Do this! ] Back to top Glen Low Guest Posted: Thu Sep 04, 2003 5:17 pm    Post subject: Re: Library/component design and implementation in C++ Quote: Yet Microsoft ATL is still stuck in macro-land. It also makes extensive use of templates, though quite a lot of that is use of the curiously-recurring template pattern. I found it a pleasure to use compared to MFC's COM support; also, much less bloated. But then, that's not saying much. I have used ATL for several small component-level projects at my day job. I like their use of templates and of course, the curiously-recurring inheritance pattern is one of my favourites. However they still insist on (1) forcing clients to use macros everywhere, and (2) foisting wizards that write code for you that you later find difficult to understand or modify. For example, all the COM_INTERFACE_MAP crap. I redid a cleaner, client-macro-free version of a COM interface map using just STL maps and virtual inheritance, it's buried somewhere in my library macstl. Quote: Java and C# proliferate without the use of generics. Yes, though there has been demand for them; they will be in Java 1.5 and Microsoft Research has added them to a research version of C#. http://developer.java.sun.com/developer/technicalArticles/releases/generics/ http://research.microsoft.com/projects/clrgen/ I have seen Java generics and a little of C# generics and they all look good, though they are runtime-based (generally it would seem on things like type erasure or similar concepts). However Java has done very well without generics and even though Generic Java existed as a proof-of-concept, somehow there wasn't much momentum behind it until C# started talking about generics too. My take on the situation -> it would be difficult to neatly retrofit into the language, especially if the VM is to be left alone, thus the long wait. Still, there's no partial specialization, which would drive the creation of a Java or C# generics-based STL. Cheers, Glen Low, Pixelglow Software www.pixelglow.com [ 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, 4, 5, 6, 7, 8  Next Page 1 of 8   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.