March 2010 - Boost - lists.preview.boost.org

GSOC Boost BigInt design issues
by swagat konchada 30 Mar '10

30 Mar '10

> >swagat konchada wrote: > > > > I had a look upon the BigInt code from the sandbox. It's > > entirely wrapped around GMP(linked to gmblib). > [snip] > > That is uncompatible with the Boost license, which hinders > > the distribution > > of a Boost_bigInt library which wholly is a wrapper around GMP. > > > > Keeping in view the above issues what should be our priority? > > Improving the > > wrapping around GMP for time being or Starting Boost's own version of > > BigInt. I see mixed reactions on this subject. My personal > > opinion is that > > suggested by Paul A. Bristow, i.e, of a Boost's own BigInt, > > which I think is in line with Boost's long term aspirations. > > > Rob Stewart wrote > The usual idea is that BigInt would provide expression templates and other > optimizations to avoid calling the back end engine inefficiently. That may > require an adapter layer between the user-visible BigInt API and the > computation engine, like GMP, at the back end, to insulate the API from > differences in the back end engines. (Then again, there may be nothing left > to a layer above the adapter layer; I'm not saying there must be three > layers.) > > If GMP is an optional batck end supported by BigInt, along with other back > ends, controlled by conditional compilation, then a BigInt user can choose > to use GMP if they like, but the licensing issue will be theirs and not > Boost's. > > HTH, > > _____ > Rob Stewart robert.stewart(a)sig.com > Software Engineer, Core Software using std::disclaimer; > Susquehanna International Group, LLP http://www.sig.com > > If we consider performance issues based on design, planning the BigInt architecture could be designed in two ways 1. 3 layers to allow optional binding to Boost's own bigint, GMP, or anyother bigint library, for the user during compile time. 2. 2 layers if we plan to limit Boost.Bigint's options to only itself and GMP. In a 3 layered architecture, 3 function calls would take place for each overloaded operator, whatever the option chosen by the user. Whereas in a 2 layered architecture, in which only GMP is an extra option, there is an equal probability of a user using Boost's own bigint or GMP and thus half of the times, 2 function calls are made and in the other half only 1 function call for each overloaded operator. This means doubling in the number of function calls. Though I realize most of the time is spent inside the function rather than in calling the functions, does this result in an overhead we should worry about? -- Problem is a phenomenon indicating lack of thought. Swagat Konchada

3 2

Library dependencies and intrer-library code reuse
by Zachary Turner 30 Mar '10

30 Mar '10

There's a lot of chatter on the list recently about how to "grow" boost. I wanted to add my own idea to the pot. Due to the scope of what I describe, I don't think it's reasonable to expect everyone to be on board with it because TBH it's a lot of work, but I'm just throwing it out there, and I think that if you forget about what would be involved in actually making it happen, it actually has many highly desirable properties. But comments would be welcome. So anyway, one of the recurring questions that pops up over and over and over is that of whether or not a library should reinvent the wheel, or use some data structure / class from some other library. On one side of the fence, there's people who say that it's often ok to reinvent the wheel inside library A if it avoids introducing a dependency on boost library B. Generally, the problem is that introducing a dependency on B a) increases compile time, because often only an extremely small component of library B is needed, but the header-onlyness of most libraries causes inclusion explosion, and b) introduces logistical problems when library B is not "permitted" for use in a certain organization. This is becoming increasingly common as certain organizations begin whitelisting boost libraries that are allowed in their codebase. On the other side of the fence, there's people who say that reinventing the wheel is almost always bad. The reasons for this are obvious and I think everyone here is well aware of the issues, so I won't repeat them. Note that I'm not taking sides here, just pointing out that the argument against reinventing the wheel is well understood. Is there a way to solve *both* of these problems at the same time? I think there is. Consider the following system: 1) Library maintenance is more flexible, and not necessarily limited to only a single maintainer or a whitelisted group of maintainers. There is another thread on this topic already, so rather than debating this particular point here, it's probably better suited in the other thread. For the purposes of this post, just assume that there is *some* system, with completely unspecified parameters, where there is *some* way for people to easily get their own code into an existing library that they are not necessarily the author / maintainer of. At the very least, *other library authors* would be given a "fast track" to submitting changes to other libraries, whatever "fast track" means. 2) All libraries would be required to support both pimpl'ed and header-only implementations. I realize this is sort of a huge undertaking and requires retrofitting tons of libraries, but I think the advantages are worth it. The users end up winning in the long run, nobody ever has to have this debate again about whether or not header only libraries are good or bad. If you want header only libraries, use #include <boost/someheader.hpp>. If you want pimpl'ed libraries, use #include <boostp/someheader.hpp>. pimpled' headers only include other pimpled headers. Also it makes the system I'm describing possible. 3) The boost build process is changed to allow additional options that either: a) automatically select all libraries for external visibility, b) select a specific list of libraries for external visibility, or c) select a specific list of libraries to disable external visibility. During the build process, what is currently known as the "boost" directory (e.g. the boost folder in #include <boost/shared_ptr.hpp>), gets a different name. Perhaps it's changed to something like <boost/internal/shared_ptr.hpp>. This happens EVERYWHERE. All boost library header and source files would need to be updated to include from boost/internal. For those libraries that were selected for external visibility, a symlink is created under the boost directory inside the internal directory. On windows a directory junction is used. Note that directory junctions are supported since Windows 2000, so I don't think this is an issue. It's very easy to create them, just that people rarely do. The end result of this is the following: 1) Users notice minimal impact. The only difference, in fact, is that if they want pimpled implementations they #include <boostp/...> instead of #include <boost/...>. This means the current behavior continues to be the default. After they build boost, the build process handles all the logistics. Everything is already in the correct structure to "just work", all symlinks have been created as described in 3) above, and the user only needs to set their include path to the top level boost directory. 2) All boost libraries can use all other boost libraries with ease, regardless of whether or not an organization's policy allows use of those libraries in user code, because internally all libraries include from boost/internal. 3) Despite the fact that all these additional dependencies that a user isn't technically allowed to be using due to his organization's policy are indeed being used, compile times can be kept sane because the user has access to pimpled libraries, and pimpled libraries only depend on other pimpled libraries, which is possible since all libraries are required to support pimpled versions. So there we go. Does this work, and if not why not? Even if we agree it's a huge undertaking, is it worth it? And if not, why not? Zach

9 16

Boost, Decoupled and Accelerated
by David Abrahams 30 Mar '10

30 Mar '10

Hi All, I know I'm not the first person to notice that, as Boost has grown, it has become harder and harder to manage, Subversion is getting slow, our issue tracker is full to overflowing, and the release process is a full-time job. For me, this situation has taken a lot of the fun out of participating. It's time to make Boost development fun again. To that end, Troy Straszheim and I have started building a system called Ryppl to decentralize development, testing, release, and installation of interdependent projects, like—but not limited to—Boost libraries. I will be holding a session **Tuesday night at BoostCon** to formally unveil this system and demonstrate it in action, and I intend to propose that Boost make the move immediately thereafter. I believe this project has the potential to change the face not only of Boost, but of open-source software in general. By the time of BoostCon, ryppl will be live, and we'll be able to use it to work on Boost. I encourage anyone who is interested in these developments to register for BoostCon immediately. If you're interested in working on development of this system, *please* send me an email; we can really use your help. Be forewarned, though: this may not operate like many other open-source projects. Given the short period of time remaining before the conference, we have little time for design discussion and debate, and we need to be maximally effective: there's simply no time to waste. You can expect me to take a “Benevolent Dictator” role, if not “For Life,” (BDFL) then at least “For the Time Being” (BDFTB). Cheers, -- Dave Abrahams Meet me at BoostCon: http://www.boostcon.com BoostPro Computing http://www.boostpro.com

12 23

[GSoC] Generic Linked List library
by Thomas Mullaly 30 Mar '10

30 Mar '10

Hi, my name is Thomas Mullaly. I am currently an undergraduate in the Computer Science department at Kent State University and I am interested in contributing a generic linked-list library to Boost as a GSoC project and I just wanted to see if there is any interest from the Boost community about having these contributed as a library. I am implementing the linked lists project from 'Elements of Programming' (ch. 12, I think), and am trying to provide lists for both constant-time size and constant-time splice. My library currently implements a number of variations of both singly- and doubly-linked lists, and I have implemented a wrapper that adds a node count and acts as a counted list. While the basic list types are already implemented, there is still a lot of work to done on them. For my GSoC project I would like to integrate allocator support into them, clean up the existing interface, and create a generic interface to the different types of lists with the help of a mentor. By the end of my GSoC project I would like to have a functional list library to contribute to Boost. I feel that this would be a valuable library to Boost because the standards implementation limits you (depending on your library vendor) to either a constant time size() function with a linear time splice operations or constant time splice operations with a linear time size() function. With my implementations a developer will be able to pick and choose whether they want a singly or a doubly linked list, or if they want to use a counted listed instead of a non-counted list. Also their is variations in memory/runtime efficiency between the different variations of the list types which a developer will be able to choose for. Currently my list implementations make heavy use of the new features available in C++0x such as rvalue references and move semantics, but, I will be more than happy to create a backport of these for the current C++ standard so they can be included in Boost. If anyone is interested in the source code for my current list implementations I will be able to provide tarballs of them.

4 3

[GSoC] Bits and Ints
by Gaurav Menghani 30 Mar '10

30 Mar '10

Hello, I am keen on working on the Bits and Ints project. However, I would be glad if someone could answer my queries. In the project description, the second line says: "Compressed value array and vector (like vector<bool> but also for multi-bit values)" 1) When we are talking of Compressed Value Array, are we looking at a functionality similar to the Deflater() class in Java, or something like Run Length Encoding / Delta Encoding? 2) I am also not sure if I get what is meant by "but also for multi-bit values", do you mean multi-dimensional bit-fields? 3) Can we discuss the proposal-drafts on the mailing list, even after the application time frame begins (March 29, 2010, 19:00 UTC)? Warm Regards, Gaurav Menghani

2 1

5 Observations - My experience with the boost libraries
by Tom Brinkman 29 Mar '10

29 Mar '10

Over the past years, every project that I have been involved with has tried and then either abandoned boost or highly restricted its use. As a long time boost user, I can't tell you how much debate we've had about this. I want boost to succeed, grow and be successful. Its sad. As someone who has used boost from the beginning, and has taken far more than I have given, here are some of my observations. Many of the reasons don't have specifically to do with boost, but more have to do with a preference for one programming style versus another. I have identified 5 key issues about why boost was abandoned. 1) Boost uses exceptions. 2) Printf versus Iostream. 3) Dependencies. 4) Research Projects 5) Macros 1) Boost uses exceptions. This issue is best illustrated when you want to just do something as simple as converting a "string" to an "int". boost::numeric_cast throws an exception, which forces you to wrap it into a try-catch block. The seemingly trivial issue of how to handle exceptions causes endless grief. It seems that developers still prefer "C" style casts. The prevailing view on projects that I have worked on is that C++ exception handling should optional and not required. int value = 0; try {value = boost::numeric_cast<int>("3.00");}catch (...){} versus int value = atoi("3.00"); I think the lesson is that there is no correct answer here. As most "C++" projects use some "C" libraries, you will need to be flexible in your approach to exception handling. A pure "C++" approach of try-catch blocks is not practical. As boost libraries tend to throw exceptions, it forces you to put try-catch blocks all over your code. Many developers will simply ignore this issue and end up with lots of code that is not exception safe. Small issue, but is undoubtably a source of a huge amount of problems. 2) Printf versus Iostream. Most developers still prefer "printf" over "iostream" style "api's". Unfortunately, Boost libraries have long ago adopted "iostream", as the preferred API style. This issue has been the source of many hours of needless debate. Pure C++ developers seem to have a aversion to "printf" style arguments. As there are very few pure C++ projects (if any), I dont hear about this as much as I once did. Most C/C++ developers prefer a printf style api. Boost needs to start providing printf style api's. Boost needs to be more practical. By not also providing printf style api's, boost is not meeting the needs of developers. Iostream may be preferable for purely academic reasons, but for practical reason, "printf" is the api style that is universally adopted. Is it boosts goal to meet only the needs of a small subset of its c++ users? If that is the goal, than it needs be clearly stated as such. As most developers have adopted a combined C/C++ style of development, a pure academic C++ style syntax is not practical for every day use. Boost needs to look at what the popular api styles are and adopt them and not be so dogamtic about its style of so-called pure "C++". 3) Dependencies. Everyone has a few favourite boost libraries. Unfortunately, because of confusing dependencies between them, its practically impossible to isolate the ones you want from from the ones you don't. What does this mean? Well, it requires the leads of a given project to create a policy about which boost libraries are approved and which ones aren't. Because of the difficulty of doing this, after a while, its can become eaiser to just say "screw it". We won't use boost at all. This issue may seem trivial, but it is the cause endless hourse of grief. It seems that once you add boost to your project, some developers feel that you should be able to use any part of boost. For obvious reasons, project leads prefer to have control over which libraries are used in a given project. Few will just give blanket approval to all boost libraries. The way boost was developed makes this practically impossible. Boost needs to advertise itself correctly and not pretend to be something its not. Its simple a place for promising research libraries to live, to get some exposure, and possibly be adopted by the larger C/C++ community. Currently, boost provides the best place to do that, but it needs to be honest with its users. 4) Many boost libraries seem more like they research projects. The prevailing view is that boost libraries push the envelope of what is possible. This is universally considered a good thing and it is always interesting to see what the best and brightest are up too. Unfortunately, boost pretends to be something else. It pretends to be a collection of libraries that meet your day to day needs. Experience proves otherwise. The practical needs of large scale active projects are in conflict. Boost is very dogmatic and encourages a particalular style of development, which is very template based. In the right hands, a heavy templated based project can work. In my experience, however, it takes many many iterations to get a template based designed library correct and usable. Many more iterations, in fact, than the equivalent non-template design. It is soo difficult, that I am very reluctant to use others template based libraries. I approach them with fear. That being said, boost::mpl and boost::spririt are wonderful accomplishments and I use them regularly. 5) Macros. Need I say more. The core boost libraries are full of macros. So much in fact, that close examination makes many of them practically unmaintainable. I know there are practical reasons to use them. The main reason being compiler inadequacies. However, many developers question the value of a cutting-edge template library that is full of macros. What is the point of having the source code, when it take an experienced developer many weeks to understand what the library is doing underneath? What is the practical value of a library that can only be maintained by the original developer(s)? This issue has surprised many, but has comes up often. If it takes more than a day to examine a libraries source code and have some idea what its doing, its very dangerous to use. This issue will cause many to avoid complex templated macro-style boost libraries. Tom Brinkman

36 161

GSOC BigInt Licensing Issues
by swagat konchada 29 Mar '10

29 Mar '10

I had a look upon the BigInt code from the sandbox. It's entirely wrapped around GMP(linked to gmblib). #inlcude<gmp.h> #ifdef BOOST_BIGINT_HAS_GMP_SUPPORT #include <boost/bigint/bigint_gmp.hpp> namespace boost { typedef bigint_base<detail::bigint_gmp_implementation> bigint; } I also read the LGPLicense which the GMP library carries with itself. This is my inference : The bigint code(from the boost sandbox) could be addressed in LGPL terms as "An Application" : which LGPL defines "as any work that makes use of an interface provided by the Library, but which is not otherwise based on the Library.", since it just uses the gmp.h interface and linking to gmplib. So the combined work, application and GMP, should be conveyed with a GNU GPL license. The Boost License allows skipping the copyright notices if only it's executables are used, unlike the LGPL which requires to distribute it's license. And there are even restrictions on "conveying the object code under terms of your(Boost') choice", in addition to other restrictions. That is uncompatible with the Boost license, which hinders the distribution of a Boost_bigInt library which wholly is a wrapper around GMP. Keeping in view the above issues what should be our priority? Improving the wrapping around GMP for time being or Starting Boost's own version of BigInt. I see mixed reactions on this subject. My personal opinion is that suggested by Paul A. Bristow, i.e, of a Boost's own BigInt, which I think is in line with Boost's long term aspirations. I hope my understanding of the issue is correct. I would be grateful to receive suggestions and comments which would help my GSOC proposal for BigInt, which I plan to submit as early as possible. -- Problem is a phenomenon indicating lack of thought. Swagat Konchada

3 2

[GSoC]
by Andrew Sutton 29 Mar '10

29 Mar '10

Just a reminder, Google Summer of Code student applications can now be submitted here [1]. The deadline for submitting an application is Apr 9. Ideas for some student projects and other information about the program can be found here [2]. Andrew Sutton andrew.n.sutton(a)gmail.com [1] http://socghop.appspot.com/ [2] https://svn.boost.org/trac/boost/wiki/SoC2010

1 0

[GSoC] Mentors group
by Andrew Sutton 29 Mar '10

29 Mar '10

All, If you are registered as a mentor for GSoC 2010, I'm going to invite you to the mentor's group/mailing list here [1]. This will be used to discuss student proposals and vote on the set of final projects for Boost. If you are already registered, then you probably won't be receiving a new invitation. This list is historically low-traffic. If you have any questions or comments, please let me know. Andrew Sutton andrew.n.sutton(a)gmail.com [1] http://groups.google.com/group/boost-gsoc-mentors

1 0

Observations - moving from boost_1_39_0 to boost_1_42_0
by Jeff Flinn 29 Mar '10

29 Mar '10

Lets not lose site of what boost is doing right. I've just updated to boost_1_42_0 from boost_1_39_0 for XCode and MSVC. 1. All but 2 previous workarounds for compiler warnings are non longer needed 2. One wrinkle in build for xcode - now require --layout=versioned 3. MSVC (distributed) build of my app using boost completes 20% faster 4. My app's full test suite runs 10% faster Thanks to all who've made this happen! If your curious, the workarounds are for the archive exceptions in portable_binary_archive, which should have been patched back in 1.40. These are actually in the examples folder, not fully supported by serialization at this time. Jeff

1 0