- Boost - lists.boost.org

[sorting] Is there still interest?
by Sam Schetterer 21 Mar '07

21 Mar '07

Well, when you timed the code, ou used std::clock. Howver, I have found that on windows, using std::time is much more accurate than clock and gives more reliable results. Also, I acknowledge that radixsort and radixquicksort might be slower than std::sort in some cass, but they are generally easier for the user to implement. Also, have you downloaded the multikey quicksort code yet? it is like radix quicksort but can be used for arrays of arrays of any type, so writing complicated and slower comparison functions are not needed for the sort to be used. For example, if you wanted to sort arrays of arrays of integers, especially if many of them are in a small range, then to write an operator<, you would have to include a for loop to compare the elements. However, multikey quicksort dows not need any complicated comparison function, so it will work really well.

3 2

[sorting] Multikey quicksort implementation
by Sam Schetterer 21 Mar '07

21 Mar '07

Earlier, I posted asking whether users would like a class that would be appended onto the end of their objects to signal that the array is over. However, that is not possible, as when appended to the end of the array, it will be treated as if it were of the same type as everything else in the array, therefore, it would be undetectable. The only other way to do this is to have users append a certain value to the end of each array that is guaranteed to be unique. This should not pose a problem for many applications, as there is almost always going to be one unique value, and in many cases, there are values that are not allowed as input into an array, making it easy to find a unique value.

1 0

Google SoC 2007 SVG Project [Boost]
by alex prince 21 Mar '07

21 Mar '07

Hey ! I hope it works this time. My message body wasn't sent last time and I was told to close the HTML editor and so I did. Here's my original question: I think I have spotted a contradiction (or at least a point that is not clear) in the SVG project.In the introduction, the author points the need to "[evaluate] a function like sin over a range and [get] a compact scalable graphic that can easily be added to documents" however in the "goal" section, the author says: "Goal: To write C++ templated functions to allow data in C++ arrays and STL containers, vector is an obvious one, to produce a file, for example, in SVG format (XML)." The latter seems to be more like simply plotting points while the first rather seems like plotting the function over a given range. I am really interested in this entry however I would like to have a really good idea of what has to be done and how I will do it before even sending my subscribtion for GSoC. Thanks a lot to all ideas and suggestions ! Alexandre P-L.

2 1

[Boost-bugs] [ boost-Bugs-1685424 ] Use of C++0x keyword as identifier
by SourceForge.net 21 Mar '07

21 Mar '07

Bugs item #1685424, was opened at 2007-03-21 13:57 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=107586&aid=1685424&group_… Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: concept_check Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Douglas Gregor (dgregor) Assigned to: David Abrahams (david_abrahams) Summary: Use of C++0x keyword as identifier Initial Comment: boost/concept/usage.hpp defines a namespace "concept". "concept" will be a keyword in C++0x, breaking this code. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=107586&aid=1685424&group_… ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Boost-bugs mailing list Boost-bugs(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/boost-bugs

1 0

Last minute intrusive review
by Daniel James 21 Mar '07

21 Mar '07

Hello, Sorry, this is review is a bit late and rushed. I feel like I'm back at college finishing a paper the night after it's due... I've read through the documentation and looked at some of the implementation (mostly the unordered containers). I think Intrusive should certainly be added to Boost. The design seems very flexible and anticipates most of the use cases that I can think of. Intrusive data structures can be useful. As well as the potential memory, performance and exception safety gains, they are perhaps a more natural way to model the relationships between entity types than containers of pointers. The documentation is good. But I think it could be improved if it showed complete code examples much sooner - at the moment it goes into the concepts in some depth before the reader gets a feel for how they fit together. I think it's possible to show how to make basic use of the simpler intrusive containers and then ease the reader into the more involved details. It's specified that islist and ilist are implemented as circular lists - this is an implementation detail and doesn't need to specified. It might also be useful to have intrusive structures without container types. Say you were implementing a smart pointer where all the pointers to the same object are linked together by a circular list. The copy constructor has a pointer would add the new pointer to list, the assignment operator would remove the pointer from its current list, and add it to the new one. All of this would be done without ever referring to a container. And it's not clear who'd be responsible for the container. This would require the hooks to be more complicated (I guess they would be something more than just hooks). And they'd probably need to use the curiously recursive template pattern. But maybe this is outside of the scope of the library. I think this may have already been mentioned by someone else, but the true/false template parameter (for constant time size) isn't very clear. In such cases, I always forget which is true, and which is false - and that's if I can remember what the parameter stands for in the first place. I think it'd be clearer to use a couple of appropriately named types. And I agree with other people's dislike of the 'i' prefix - it's already used in the string algorithms for 'case-insensitive'. Also ilist and islist look a little too similar. I had a quick look at the tests and it looks like there could be some more (which is of course always true). Exception tests could be useful where needed (Boost.Test has a pretty good exception testing framework, but it's not documented). Finally, something which I haven't spent enough time on. And it's late. So I'm probably saying something really stupid. But basic containers seem a little too complicated to define. Especially for dependant types inside templates. Something like: typedef islist< islist_base_hook<my_tag>::value_traits<MyClass> > BaseList; Could perhaps be: typedef islist<MyClass, islist_base_hook<my_tag> > BaseList; With a third parameter that could used to specify an alternative value_traits when needed. Which I think fits in better with the typical STL way of doing things. It might be possible to do something simpler than this for some special (but common) cases - but the only ones I've thought of so far are problematic. It does seem odd to have to sepecify 'islist_base_hook' for 'islist' - I'd expect that to be the default. I guess to elide that there would need to be some mechanism to distinguish between tags and hooks - allowing you to use just the tag. A bit iffy. Also, a default tag would be useful. Most of the time a container will only have a single hook of a certain type and declaring a tag in this case is unnecessarily verbose. Daniel

2 1

[Boost-bugs] [ boost-Patches-1685367 ] [lambda][utility] Make lambda support result<>
by SourceForge.net 21 Mar '07

21 Mar '07

Patches item #1685367, was opened at 2007-03-21 12:54 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685367&group_… Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Daniel Walker (daniel_walker) Assigned to: Nobody/Anonymous (nobody) Summary: [lambda][utility] Make lambda support result<> Initial Comment: Apply with 'patch -p0 < djw_lambda_result.patch' from the boost root directory. This patch also defines macros in boost/utility/result_of.hpp that generate nested result<> members for user defined functions of various arities. See discussion here http://tinyurl.com/2jgtld. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685367&group_… ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Boost-bugs mailing list Boost-bugs(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/boost-bugs

1 0

[Boost-bugs] [ boost-Patches-1685360 ] [parameter] fix operator|| for lazy binding
by SourceForge.net 21 Mar '07

21 Mar '07

Patches item #1685360, was opened at 2007-03-21 12:47 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685360&group_… Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Daniel Walker (daniel_walker) Assigned to: Nobody/Anonymous (nobody) Summary: [parameter] fix operator|| for lazy binding Initial Comment: Apply with 'patch -p0 < djw_lazy_binding.patch' from the boost root directory. See discussion here http://tinyurl.com/26kwnh. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685360&group_… ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Boost-bugs mailing list Boost-bugs(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/boost-bugs

1 0

[Boost-bugs] [ boost-Patches-1685354 ] [utility] Make result_of handle lambda expressions
by SourceForge.net 21 Mar '07

21 Mar '07

Patches item #1685354, was opened at 2007-03-21 12:41 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685354&group_… Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Daniel Walker (daniel_walker) Assigned to: Nobody/Anonymous (nobody) Summary: [utility] Make result_of handle lambda expressions Initial Comment: Apply with 'patch -p0 < djw_result_of.patch' from the boost root directory. See Discussion here http://tinyurl.com/2h3g7n. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685354&group_… ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Boost-bugs mailing list Boost-bugs(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/boost-bugs

1 0

[intrusive] Review
by Guillaume Melquiond 21 Mar '07

21 Mar '07

* First, a few points that need to be addressed, or at least commented upon, in my opinion. The member hooks are parametrized by the class name, and I can't see why it is needed. As the matter of fact, the base hooks don't need the type, so the member hooks should not either. In my opinion, it is just a waste of compile time and of executable space (think run-time or debug type information). Moreover, it would make the interface simpler and remove a discrepancy between base and member hooks. This is current situation: _base_hook<Tag>::value_traits<Class> _member_hook<Class>::value_traits<&Class::Hook> Notice that the Class parameters do not appear at the same place. After removing the first template parameter from _member_hook, the traits would be accessed by: _base_hook<Tag>::value_traits<Class> _member_hook<>::value_traits<Class, &Class::Hook> Now the situation is uniform, the Class parameter always appear as the first parameter of value_traits. Speaking of the interface, half of the template parameters of the intrusive container are used for specifying the type of the variable storing the container size. This does not need that many parameters, as they don't provide orthogonal features. So just one is enough: template < class Traits, class Cmp, class SizeType = std::size_t > Then you specialize your containers, so that the size is not stored when the type is void. If you think void is not descriptive enough, you can replace it with a dummy type called no_constant_time_size, or whatever. In the hooks, the assignment operators failing when the destination node is already linked feels wrong. I understand what the rationale for iset could be: assignment may change ordering of nodes. But for ilist, there is no ordering involved, so assignment should be allowed. And for iset, it should fail only when the ordering is broken, but I don't think it is feasible to verify this condition; so it might be sensible to allow it there too. Concerning the advanced insertion functions, I'm not sure if they are really useful. It seems to me like the insert_check/commit functions for iset are just redundant with the usual way of doing it in the STL: calling lower_bound, checking the iterator, and then hinting the insertion into the set. Especially since you already provide an optimized lower_bound function, which can be used without creating the value. About auto-unlink hooks, you say that they must be used carefully because "removing the object is thread-unsafe". I'm not sure to understand what you mean. Even for manual-unlink, it is unsafe to remove an object if several threads are using the same container. So is there a particular reason to mention it here? Concerning assign and copy of intrusive containers, I agree that it makes little sense. But only if both containers have the same type. You could provide template operators in the case types differ but value_types are the same. Notice that I'm talking about doing the equivalent of "set2(set1.begin(), set1.end())", so the cumbersome clone_from does not work here. The ilist class provides an assign_and_destroy function. The name is misleading, as the function will destroy the old elements and then assign new elements. In comparison, all the other _and_destroy functions delete the elements acted upon, while assign_and_destroy deletes unrelated elements. Perhaps, this function could be removed, as it does nothing more than just clear_and_destroy then assign. The irbtree class is a nice thing to have, so it shouldn't be hidden as a detail, especially as its interface is similar to the ones of the other containers. Typo in advanced_lookups_insertions.html, "a name it's enough". * Now for the review itself. - What is your evaluation of the design? The concept of intrusive containers is sound, and the library is really generic. Some bits of the interface may require a bit of tuning (see the first two points above). - What is your evaluation of the implementation? The implementation looks good. I especially appreciated the fact that the default classes were relying on void* to reduce code duplication. - What is your evaluation of the documentation? It contains all the needed information, but it is a bit complicated to get at it. In particular, the class references should be directly accessible from the main page. The introduction could also benefit from explaining what an intrusive container is and why to use it. - What is your evaluation of the potential usefulness of the library? It can be extremely useful for some algorithms, when elements have only one owner but you want them to be part of several structures. - Did you try to use the library? With what compiler? Did you have any problems? I exercised the library by building a non-intrusive multiset upon it. I compiled with GCC 4.1. The implementation was a bit slower than the original std::multiset, but nothing too worrying. I didn't encounter any problem: thanks to the intrusive library, the implementation was straightforward. - How much effort did you put into your evaluation? A glance? A quick reading? In-depth study? In-depth study. - Are you knowledgeable about the problem domain? My code is a heavy user of intrusive lists. I never went as far as building an intrusive set though, but I do have some use for it. - Do you think the library should be accepted as a Boost library? Yes, I vote for inclusion. But I would really like the two details I mentioned about the interface to be solved before it is engraved in stone. Best regards, Guillaume

5 8

[Boost-bugs] [ boost-Patches-1685351 ] Detect nested template members with has_template_xxx
by SourceForge.net 21 Mar '07

21 Mar '07

Patches item #1685351, was opened at 2007-03-21 12:38 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685351&group_… Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Daniel Walker (daniel_walker) Assigned to: Nobody/Anonymous (nobody) Summary: Detect nested template members with has_template_xxx Initial Comment: This patch is for Boost.MPL. Apply with 'patch -p0 < djw_has_template.patch' from the boost root directory. See discussion at http://tinyurl.com/35x5z5. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=307586&aid=1685351&group_… ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Boost-bugs mailing list Boost-bugs(a)lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/boost-bugs

1 0