- Boost - lists.boost.org

IndexedSet Review: unified proposal for naming
by Joaquín Mª López Muñoz 01 Apr '04

01 Apr '04

I don't think any more people will join the review, so I've collected all the contributions regarding naming issues and tried to come up with a "unified" naming scheme. As it was expected, suggestions vary wildly, and the choicespan is growing as the review advances, so maybe this is a good time to try to reach a compromise. Please note everybody will have a favorite alternative, but the idea is that this proposal at least satisfies everybody and is abhorred by nobody. I'd appreciate if you take a look at the proposal and either accept it or provide strong argument against it (or against parts of it). It would help if further discussions conform more or less to the structure of this proposal, so as to ease gathering results later. Also, I'm not the review manager and it is not for me to make the final decision on this and other issues, but I think some preprocessing work could be beneficial. So: NAMESPACE * Proposal: boost::container::multi_index * Rationale: The name suggests what the library is about and does not clash with the name of the container. Preferred to multiindex as English seems to like the former better. In favor of multi_index: Rob, Pavel (underscore-less variant), Arkadiy (for the container instead), Jeff, Thorsten, Paul, Matthew (only if pushed into boost::container). Alternatives: non-descriptive name (Tom), multi_indexed_containers. * Why inside boost::container? Seems like there's a general move to 2nd level namespaces, and this in particular saves us from repeating "container" in the library namespace itself. If anything, being the only library inside boost::container does not hurt. I've scanned the posts and these are the reviewers' opinions about this subject matter: In favor of pushing into boost::container: Matthew, Joerg Against: Dave Not sure: Jeff Don't care: Gary, Darren CONTAINER NAME Proposal: composite_container. Rationale: It more or less suggests what it is. "index" does not appear in the name, but to compensate it'll show anywhere else (the namespace and the parameters like index_list) so the index concept is ubiquitous. It does not use "set" as composite_container in general is not a set. It matches the concept it'll model in the future (when this section is written.) Alternatives proposed by reviewers: indexed_set, multicontainer, indexed_table, table, indexed, composite_set, set (distinguised from std::set by the namespace). LIFTING TO NAMESPACE BOOST * Proposal: container lifted only one level: boost::container::composite_container * Rationale: Lifting to boost seems too long a jump, and lifting to both places is overkill. LIBRARY NAME * Proposal: Boost Multi-index Container Library * Proposal for short form: BMCL * Rationale: The name is descriptive enough, and gives room for future containers to be added here (indexed maps, the infamous bimap, etc.) REGULAR INDICES * Proposal: Name changed to "ordered indices". * Rationale: The name is far more descriptive. When first met, "regular" means nothing. Allows for a nice solution to the clash with std::unique (see below). INDEX NAMES * Proposal: "unique" and "non_unique" changed to "ordered_unique" and "ordered_non_unique". * Rationale: No one opposed and some liked it. Solves the clashing with std::unique. Extends nicely to hashed indices (unordered_unique etc.) It is more descriptive. In favor: Gary, Dave. Rob dislikes the unique/non_unique suffix, but alternatives do not look better IMHO. NTH_INDEX_TYPE AND FAMILIY * Proposal: nth_index<N>::type index<Tag>::type nth_index_iterator<N>::type nth_index_const_iterator<N>::type index_iterator<Tag>::type index_const_iterator<Tag>::type * Rationale: It eliminates the dreaded "_type" suffix. It does not clash with inherited "iterator" and "const_iterator" types. Two reviewers loved it (Darren, Dave), the rest said nothing about it. UPDATE MEMFUN * Proposal: Change to "replace". * Rationale: It more faithfully describes the semantics of the operation. One reviewer in favor (Dave), plus Pavel and myself. HEADER ORGANIZATION * Proposal: follow a fine-grained policy, possibly with "grouping" headers (e.g. for key extractors.) * Rationale: Seems like good style. Two reviewers favored this approach (Dave, Gary). The rest didn't care/say anything. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

5 5

Re: [boost] Re: Formal Review: Indexed Set
by JOAQUIN LOPEZ MU?Z 01 Apr '04

01 Apr '04

Hi Fredrik, thank you for the reviewing effort! ----- Mensaje original ----- De: Fredrik Blomqvist <fredrik_blomqvist(a)home.se> Fecha: Martes, Marzo 30, 2004 11:40 pm Asunto: [boost] Re: Formal Review: Indexed Set [...] > * What is your evaluation of the design? > I don't feel quite comfortable with the assymetry of index-0 not > needing a > get<> accessor. I'd prefer forcing the use of get<> everywhere but > I guess I > can live with it if thats decided. More real-world usage should reveal > wether a substantional amount of use-cases naturally benefit from > this bias. Others have also expressed dislike for this feature. Admittedly, it is easier to go the other way around: releasing first without default get<0> and gaining experience to see if it's worth adding on a later release. OTOH, I find the feature useful... I don't know. > > I also second the requests for lifting an implementation of a > bidirectional_map/set to official status. But I don't expect nor > demand this > to be added before acceptance, I'm fine with it being a reasonable > prioritized roadmap feature. Count on it. I'm emotionally attached to bimap, as this container was that started it all. > > Serialization support is also high on my wish-list (but with same > remarks as > above). It is already in the todo list. I'll try to reserve some time to enter into the serialization review and use this case as a testbed. > * What is your evaluation of the implementation? [...] > -I think it would be better if the iterators (index_iterator.hpp) were > generated using boost.iterator_facade/adapter instead of the > boost.operatorshelper. This would make it easier to adapt and for > example directly support > the new iterator concepts. Ummm... will look at it. > -This is more of a "feeling", but could perhaps (parts of?) the > 'member'functionality be refactored to use boost::bind/mem_fn code > instead? I don't think so. member<> and the other key extractors are polymorhic, meaning that they accept not only objects of type T, but also reference_wrapper<T>'s, pointers to T and some more. I cannot figure out how to do this with boost::bind. > * What is your evaluation of the documentation? [...] > Some nitpicks: > -In the example at the end of the tutorials section ('Projection of > iterators') > I think it would be clearer if it would use equal_range instead. You cannot! I was bitten by this at my first try to write the snippet. Suppose we do an equal_range() on "sister" and get the range [it0,it1): it0 will point to a sister and *it1 will point one-past the last occurrence of "sister". Imagine now that we are prepending "younger" instead of "older" and that *it1 is "zoe". Then while(it0!=it1;++it0) { // get it2 from it0 by means of project() tc.insert(it2,"younger"); } kah-boom! Now, *as seen by index #1*, "younger" gets between the last occurrence of "sister" and "zoe" (alphabetical order). So it1 will be pointing off beyond the "sister" range (in fact, an infinite recursion ensues.) I could have fixed this by considering (--it1) instead, but I thought the way I settled on was clearer for explanatory purposes. > > -I find the coding style to be slightly too compact. No space > after ',' and > '&' etc makes it more difficult to quickly browse. Why not just > follow the > semi-official boost coding guidelines? (currently found in the > file area) No particular reason for the style, I've been coding like this for 10 yrs. Other reviewer requested that lines did not exceed 80 chars, which I've just done. But inserting the missing blanks is a pain in the ass, and I'll probably start coding again like before as soon as I forget about it. > > -Several of the code example in the tutorial section could be slightly > reformatted to use less horizontal screen space (see for example > the 'Key > extraction' part). > I'd rather scroll vertically than horizontally! Simply pulling in > some of > the end of line comments would fix most of it. > Previously reported request, it's in my todo list. > * What is your evaluation of the potential usefulness of the library? [...] > * Did you try to use the library? With what compiler? Did you [...] > Hope to be able to use it in larger scale shortly. Please do it and report your experience. > * How much effort did you put into your evaluation? A glance? A quick > reading? In-depth study? [...] > * Are you knowledgeable about the problem domain? [..] > * Do you think the library should be accepted as a Boost library? > > Yes, I vote for acceptance. Well, thank you! I hope you can find this lib of some use in the future. Regards, Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 1

problem with global mutex object on windows
by Stefan Seefeld 01 Apr '04

01 Apr '04

hi there, we'v run into a problem with the boost_thread mutex class, when used as a global object. Imagine a singleton class as class Singleton { public: static Singleton *GetInstance(); static void Destroy(); private: struct Guard { ~Guard() { Singleton::Destroy();} }; static Singleton *instance; static Guard guard; static boost::mutex mutex; }; where the internal access to 'instance' is serialized by means of the mutex (inside methods GetInstance() and Destroy()). It is the Singleton::Guard destructor that is responsible to free the instance (if it isn't destructed already). It, too, uses the mutex to protect access. In a specific context on windows xp it happened that the mutex variable was destructed before the guard, leading to a global application lock-up when the Singleton::Destroy() (called from the guard destructor) attempts to aquire the lock. Is there a way around this dilemma ? During debugging I set the 'm_mutex' member of the mutex class (which is a CRITICAL_SECTION) to zero right after deletion, which worked, i.e. there was no lock-up. I'm not sure though that this is the correct thing to do. Any suggestions ? Thanks, Stefan

2 1

Formal Review: Indexed Set
by Pavel Vozenilek 01 Apr '04

01 Apr '04

The review of Indexed Set library, written by Joaquín M López Muñoz starts today (March 20) and runs for 10 days. Latest version (9.4) of the library is available on: http://groups.yahoo.com/group/boost/files/indexed_set.zip (328 kB) What it is? indexed_set is generic, STL compliant container providing one or more views on its data. These views allow to use data sorted according to different criteria. Change in container data is automatically propagated into all views. How many views and what ordering is applied is specified in declarative way, as template parameter. What it is good for? - it can replace multiple collections of data and hand-written code keeping them in sync. - it can replace standard containers (as std::set or std::list) if these may not meet future reqirements. - it can serve as basic for other libraries. For example RTL (relational tables library) may use indexed_set to hold its data. Other features of indexed_set: - sorting can be performed on result of member function (similar to calculated indexes in RDBMS). - indexed_set can be used as 'wrapper' over other data structures, e.g you can see data in std::vector as being sorted in set-like structure (but without structures being synchronized). - many, many others ... Future development: - planned functionality is described on its own page in documentation. - indexed_set may also implement access to its data in the same way as std::deque, with random O(1) access History of the project: - it started as 'bimap' library more than year ago, - some two dozen iterations were released, - several Boost people expressed their interest or did suggest features. Boost-wide reusable parts of the library: - indexed_set includes (nearly verbatim) copy of Andrei Alexandresu ScopeGuard. It is intended as stop-gap until Boost wide library emerges. - maybe auto_space.hpp can be liften into utilities. Portability: - see document page for list of compilers and their issues - Can someone try to run tests under release mode of VC 6.5? We got different results with different machines. Word from the autor: "Thank you for reviewing indexed_set. There are several points, mostly concerning naming, that I'd like to be discussed during the review. You can find these at the URL http://groups.yahoo.com/group/boost/files/indexed_set_review_notes.html If you evaluate the library and deem it acceptable for Boost, please walk the extra mile and express your opinions about the issues brought forward in the review notes. Naming issues are quite sensitive, and I'd like to reach as general a consensus as possible." Your review comments are welcomed. Found problems and suggestions for documentation are highly sought. Requests for new features requests may shape way indexed_set will evolve in future. As naming questions can generate lot of mail-list traffic, I'll start separate thread to deal with them. /Pavel

12 36

geometry2d: Whats the status of geometry2d in the boost vault?
by Tom Brinkman 01 Apr '04

01 Apr '04

There is an interesting class on the boost vault called geometry2d. What is the status of this class. Its related to the proposed region class. http://groups.yahoo.com/group/boost/files/geometry2d __________________________________ Do you Yahoo!? Yahoo! Finance Tax Center - File online. File on time. http://taxes.yahoo.com/filing.html

3 2

RE: [boost] Re: Re: Region Library?
by Hanson, Iain 01 Apr '04

01 Apr '04

> From: Fernando Cacciola [mailto:fernando_cacciola@hotmail.com] > Sent: 01 April 2004 17:18 > > > "Hanson, Iain" <iain.hanson(a)csfb.com> escribió en el mensaje > news:4DC0299EAC0ED311B7580090274CF0640B690B1C@slon12306.csfp.co.uk... > > [ snip ] > Though I would certainly found such a class useful, I'm not > sure if its > place is Boost. > Seems too domain-specific to me. > Why not sourceforge, or maybe flipCode or any of the more > game/graphics-centric repositories. Unfortunately, most of them don't write generic code and I think that you are being a little pessermistic with regard to applicable domains. However, even if this was their only domain of applicableity, I'm not sure that it is any more restrictive than octonion & quaterion that are already in boost. /ikh ------------------------ #include <std_disclaimer> Everything following this line has been provided courtsey of my ISP and I have no control over it. ============================================================================== This message is for the sole use of the intended recipient. If you received this message in error please delete it and notify us. If this message was misdirected, CSFB does not waive any confidentiality or privilege. CSFB retains and monitors electronic communications sent through its network. Instructions transmitted over this system are not binding on CSFB until they are confirmed by us. Message transmission is not guaranteed to be secure. ==============================================================================

1 0

[optional] proposal for extension
by JOAQUIN LOPEZ MU?Z 01 Apr '04

01 Apr '04

In the rationale of Boost.Optional, optional<T> is proposed as a convenient replacement of the time-honored practice of returning a std::pair<T,bool> with the bool indicating whether the T object is meaningful or not. Notably, this is precisely what some STL containers' memfuns do. So why not augment optional<T> with the following: optional<T>::optional(std::pair<T,bool> const&); and the obvious semantics? This would allow for instance to use optional<> like this: std::set<int> s; ... boost::optional<int> oi=s.insert(10); // oi is initialized iff insertion took place. Comments? Apologies if this has been discussed previously. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

6 5

RE: [boost] Re: Region Library?
by Hanson, Iain 01 Apr '04

01 Apr '04

> From: Allan Odgaard [mailto:ML@Top-House.DK] Sent: 30 March 2004 22:25 > > > Yes -- although I personally do need it even when I do not write > operating systems ;) and thus my question really was if there is any > general interest in such a library, since I did exhaustive searching > and was only able to find such region stuff implemented > publicly in the > GDK/GTK library, which is ANSI-C AFAICT, and that led me to the > conclusion that either there is no real use for this stuff > outside GUI > kits, windowing systems, and 2D games or efficient region "logic" is > often skipped for ad hoc workarounds like using bounding boxes or > refreshing more than necessary -- in which case there really ought to > be a library. > I think that there really is a need for such a library. From what I have seen, many libs that need this either use ad hoc solutions as you suggest above, or they re-invent the wheel and the solution they come up with is intimately bound up with their problem space and therefore not easily re-useable. /ikh ------------------------ #include <std_disclaimer> Everything following this line has been provided courtsey of my ISP and I have no control over it. ============================================================================== This message is for the sole use of the intended recipient. If you received this message in error please delete it and notify us. If this message was misdirected, CSFB does not waive any confidentiality or privilege. CSFB retains and monitors electronic communications sent through its network. Instructions transmitted over this system are not binding on CSFB until they are confirmed by us. Message transmission is not guaranteed to be secure. ==============================================================================

2 1

Re: [boost] geometry2d: Whats the status of geometry2d in theboostvault?
by Tom Brinkman 01 Apr '04

01 Apr '04

>> SFINAE makes it possible, and we have something proprietary along these lines here at work. I'd by happy to elaborate more on specific questions if you are interested to know more. Yes, sounds interestion. While I'm sure that your concerns are valid about this library in its current state, i've found this library to be good enough. It may not be perfect, but its a better rectangle class than any i've seen. I think that we should get it reviewed by others and try to get it added to the main boost cvs. When an improved geometry class comes along, we could add that one to boost as well. Geometry2d is not a complicated library, as it consists of only one header file, and would be easy enough to just add it to the utility folder. Just my thoughts, Tom brinkman __________________________________ Do you Yahoo!? Yahoo! Finance Tax Center - File online. File on time. http://taxes.yahoo.com/filing.html

3 2

C++ subranges
by Brian Martin 01 Apr '04

01 Apr '04

Hi Brian Martin wrote: > I have a suggestion for a simple Boost feature - subranges. [...] > The big question is, do any of you think this is a useful enough feature? Maciej Sobczak wrote: >Yes, it is. >What's more, the biggest gain (in my humble opinion, of course) is not >the bare fact that you can have range checking at run-time, but that it >is possible to provide range checking at *compile-time*, thus >strengthening the type system. There are some interesting ideas in your code. The current version of my subrange class does not trap potential compile time errors. Your technique of using template specialisation for this is a good idea but unfortunately the following gets past my (Borland) compiler: range<0, 200> k4(500); and it would be nice to catch this. Also, I don't see how the following can ever be trapped by a compiler: range<0, 100> i(4); i = 98; i++; ++i; surely that is a runtime thing. One possible idea for catching the first one is to redefine the template as follows: template <long L, long U, long V> class range { long v_; public: range() { static_assert<L <= V && V < U>(); v_ = V; } }; I've tried this on a simple example and it seems to work. Unfortunately, what you pass to V has to be a constant expression. Thus this won't compile at all: const long x = 2; range<0, 10, x> my_range; Even though x is in range and const. I suppose it's a const "variable". Drat. Your class doesn't have the concept of a "base type". It's a template restricted to longs. Class subrange allows restriction of any type that supports the appropriate numerical operators (even classes - admittedly a bit dubious). This is similar to the Pascal facility where you can have subranges on any ordinal type (i.e. integral numerical, character or enumeration) e.g. type LowerCaseChar = 'a'..'z'; type UpperCaseChar = 'A'..'Z'; type DaysOfTheWeek = (Sunday, Monday, Tuesday, Wednesday, Thursday, Friday, Saturday); // equal to a C++ enum declaration type WorkingDay = Monday..Friday; // For those who don't work weekends! I know that C++ does implicit conversions from say char to int etc. e.g. typedef range<'a', 'z'> lowercase_char; but typedef subrange<char, 'a', 'z'> lowercase_char; is a bit more "self documenting". The specification of the base type char makes it clear we're storing characters and not longs. I'll have another think about some of the compile time check issues. Regards, Brian Martin _______________________________________________ Unsubscribe & other changes: <http://lists.boost.org/mailman/listinfo.cgi/boost>

1 0