- Boost - lists.boost.org

Re: [boost] Re: compressed_pair requirements?
by JOAQUIN LOPEZ MU?Z 27 May '05

27 May '05

----- Mensaje original ----- De: Peter Dimov <pdimov(a)mmltd.net> Fecha: Viernes, Mayo 27, 2005 7:47 pm Asunto: Re: [boost] Re: compressed_pair requirements? > Howard Hinnant wrote: > > > 2. Uninitialized is better because it is much faster, > especially if > > you have containers of such elements. You can always explicitly > > initialize if you want. > > Technically... you can't put an uninitialized (non-character) > scalar in a > container; copying an uninitialized scalar is undefined behavior. > A similar issue also renders the following snippet illegal (unless someone has filed a DR): typedef set<int> int_set; pair<int_set::iterator,int_set::iterator> p; According to 20.2.2.2, the default ctor of std::pair works as if implemented like pair():first(int_set::iterator()),second(int_set::iterator()){} and int_set::iterator(), being a singular value, cannot be copied. A technical nitpick save for the fact that I actually was once static asserted by an overzealous checked STL implementation (don't remember which, I think it was Comeau). IMHO, std::pair default ctor should be defined as pair():first(),second(){} Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

2 1

[property_map] Template parameter disorder
by Cromwell Enage 27 May '05

27 May '05

According to <http://www.boost.org/libs/property_map/iterator_property_map.html>, ValueType and IndexMap should be the third and second template parameters, respectively. Should <boost/property_map.hpp> be fixed to match its documentation? Cromwell D. Enage __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

1 0

[test lib] request for new features
by Thorsten Ottosen 27 May '05

27 May '05

Hi Gennadiy, I would like to see these functions added to the test library: BOOST_CHECK_NOT_EQUAL(x,y) BOOST_CHECK_NOT_CLOSE(x,y) and any similar functions that makes sense. I don't see how we can get the variables printed in the failed test without these functions. Thanks Thorsten -- Thorsten Ottosen ---------------------------- www.dezide.com http://www.cs.aau.dk/index2.php?content=Research/mi www.boost.org www.open-std.org/JTC1/SC22/WG21/

3 7

build from cvs, what is ICU_PATH?
by Neal Becker 27 May '05

27 May '05

I'm trying to build from cvs on Linux. Here's what I get: PYTHON_ROOT=/usr PYTHON_VERSION=2.4 bjam -sTOOLS=gcc stage ****************************************************** Building Boost.Regex with Unicode/ICU support disabled. HINT: define the environment variable ICU_PATH to point to the root directy of your ICU installation if you have one. Couldn't find utypes.h in ****************************************************** locate utypes.h /usr/include/schily/utypes.h Is this what it wants? How to I use it?

3 2

RE: [boost] Re: Customer Friendlier Boost Installation
by Brian Braatz 27 May '05

27 May '05

> -----Original Message----- > From: boost-bounces(a)lists.boost.org [mailto:boost-bounces@lists.boost.org] > On Behalf Of Loïc Joly > Sent: Thursday, May 26, 2005 2:23 PM > To: boost(a)lists.boost.org > Subject: [boost] Re: Customer Friendlier Boost Installation > > Paul A Bristow a écrit : > > > Recently we had a flurry of discussion of why Boost is not more > widely used. > > > > I have used some bits of Boost happily for years, but recently I felt > I had > > to use it 'properly' so concluded it was time I used the bjam build > system - > > something I had felt looked a bit clever. > > > > > <SNIP> > > * Finally, installing boost takes longer than installing VC++... Is > building from the source such an important point ? Would'nt it be > possible to provide pre-built libraries for the 3 or 4 main compilers on > windows with each release ? The whole package, zipped, is 80mb on my VC8. > > -- > Loïc [Brian Braatz Writes:] Thank you for also suggesting that. Somewhere, someplace, an angel just got his wings.. :)

2 1

Requests for changes in 1.33 related to serialization
by Neal Becker 27 May '05

27 May '05

I want to suggest 2 changes that would be helpful at least to me, that relate to serialization. I use serialization with mersenne_twister and python. To support this 2 changes are needed: 1) serialization must be build as a shared library. I know this doesn't work on all platforms, but I believe it works on at least some, and therefore should be supported. 2) Please add support for serialization to mersenne_twister.hpp. All that's needed is this: template<class Archive> void serialize(Archive & ar, const unsigned int version) { ar & x; ar & i; }

6 9

Re: [Windows MSI Installer] - Using WIX
by Peter Foley 27 May '05

27 May '05

Hi, To be perfectly honest I haven't really looked for anything else and unless people can come up with a reason for not using WIX it is likely the one I will use. Why? Mainly for 2 reasons: 1. This is the one I am most familiar with (ie been reading about it for a long time) 2. I have been meaning for a long time to actually try using WIX. I guess this is my chance =p. Now does this mean it is the right tool to use for Boost? Hell no!!! But as I said unless there are good reasons for not using it I will probably just stick with this one. Although if other people have already started something else I am happy to contribute =p. Peter. ""Lars Gullik Bjønnes"" <larsbj(a)gullik.net> wrote in message news:<m3mzqh6je5.fsf(a)tyfon.gullik.net>... > "Peter Foley" <pjfoley(a)bigpond.com> writes: > > | Hi, > | > | I am in the process of looking at how feasible it would be to use WIX to > | create an MSI installer package for Boost for the Windows Platform(s). > | > | Now bearing in mind that I have never used WIX before and have been out of > | the programming game for a while I think I would like to help/run with this > | task. > > I know that several Open Source projects use the NSIS installer. > > You should perhaps have a look on that one as well. > > http://nsis.sourceforge.net/ > > -- > Lgb > > _______________________________________________ > Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost >

1 0

Re: [Windows MSI Installer] - Using WIX
by Peter Foley 27 May '05

27 May '05

Hi David, Comments inline. "David Abrahams" <dave(a)boost-consulting.com> wrote in message news:<ufyw9c5qm.fsf(a)boost-consulting.com>... > "Peter Foley" <pjfoley(a)bigpond.com> writes: [SNIP ...] > > * Install boost using an installer application > > * Detect necessary installation information > > * Provide "header only install" or header and lib file install" > > * Install foo library and dependencies and that is it > > * Configure various IDE's to support using boost > > * Install and register an MSDN Library of the boost documentation > > Did you miss my list or did you dislike everything on it? > http://article.gmane.org/gmane.comp.lib.boost.devel/124787 No I didn't miss it, I receive the list in Digest mode which makes it hard to find stuff sometimes (especially with this thread 150 replies). That list I created was from memory I am happy to incorporate your suggestions. [SNIP ...] > > > 1. Can we write a program that could work out if a library is header > > only? (generating a list of all these files) > > What do you mean by "work out?" Oh, I think you mean "decide." At > first I thought you meant "function properly." > > Yes, we can. What I am driving at is could we create a script/tool that would generate a list detailing these files and write it to a file that can be easily re-interpreted (ie output in XML or structured text). Maybe something like: [Library Name] [File Name] Foo Library foo.hpp During the build process this file will be read and incorporated into an XML document that the Candle (WIX Compiler) and then Light (WIX Linker) will generate the MSI from. > > > 2. Can we write a program that could work out if a library needs to be > > compiled as a lib file? (generating a list of all these files) > > Isn't that just the inverse of question 1? I guess so =p. Thinking about it the same tool described above could include this information. So modifying the Columns: [Library Name] [Header Only] [File Name] Foo Library YES bar.hpp The output will obviously be more complicated then this (to build the library you need the Jam file and anything else you usually would need). > > > 3. Can we write a program that could work out the dependencies between > > different libraries? (generating a list of all the dependencies) > > bcp I have not used bcp so excuse my ignorance if it already can do this. Can you configure it to run across the whole boost source tree and spit out an easily re-interpreted file listing the dependencies between all libraries? > > > 4. Is there a number of files that you would consider to be core boost > > files (ie something that the majority of boost libraries/users would use)? > > Yes, but why does that matter? If we allow people to limit how much > they install, they'll be offended if they get something they don't > need, even when it's in the "core." I guess I need to elaborate here. This isn't necessarily for the end users this is to hopefully simplify the dependency graph. Ie I am guessing a majority of libraries would use smart_ptr? Are there other libraries like this? To give you an indication on how I think an end user would use the installer (bare in mind this is pie in the sky stuff till I find out if it is possible =p) 1. Download MSI package from boost website, which will include - Bjam (pre-built) (zipped currently 83k) - Boost Sources (zipped currently 19MB) 2. Run the MSI and follow the bouncing ball (note this is not set in stone, just an idea of how I would love to see it) - View the boost License file(?) - Check for various Toolsets - Choose parts of boost to install, broken down by: (Note: Would need dependency information to break it down to this level) * Header Only libraries + Header A + Header B + Header C + Header D * Compiled libraries + Lib A + Lib B + Lib C + Lib D * Documentation + All (Default) + Selected header/lib's documentation * Rest of the support files needed for boost development (At a minimum I would default the installation to Headers, All documentation. But this is open to discussion and would be easy to chop and change) - Choose boost installation location - Choose the toolset - (Display if compiling lib's) Configure any additional options Bjam would use - Display a summary of the tasks about to be preformed 3. Once they click on finish the installer will - Extract the Header files and documentation to install directory - Invoke the command line and run the Bjam command to create the lib files - Update the toolset to use the boost files As long as everything works out the process to create a new MSI file for a new boost version for windows would be: 0. Get the stable boost branch from CVS 1. Generate header/lib XML file 2. Generate dependencies list 3. Process these files into the WIX XML document 4. Compile the WIX XML document 5. Link the WIX object files 6. Upload MSI Package to sourceforge This could be down by Script, Bjam, Make or by hand. Peter.

1 0

RE: [boost] Re: Wave Bjam building problems\Joel de Guzman-didyoueverfind a solution?
by Brian Braatz 27 May '05

27 May '05

> -----Original Message----- > From: boost-bounces(a)lists.boost.org [mailto:boost-bounces@lists.boost.org] > On Behalf Of Hartmut Kaiser > Sent: Thursday, May 26, 2005 9:58 AM > To: boost(a)lists.boost.org > Subject: RE: [boost] Re: Wave Bjam building problems\Joel de Guzman- > didyoueverfind a solution? > > > David Abrahams wrote: > > > > I am now trying to build the quick start with the VC71 IDE > > > > > > My current problem is this link error: > > > Linking... > > > LINK : fatal error LNK1104: cannot open file > > > 'libboost_wave-vc7-mt-gd-1_32.lib' > > > > > > Build log was saved at > > > > > "file://c:\data\Irx\Irx\Arx\Main\Source\Common\Boost\V_1_32\libs\wave\ > > > sa > > > mples\quick_start\Debug\BuildLog.htm" > > > WaveQuickStart - 1 error(s), 0 warning(s) > > > > > > > > > The LIB I have available is 'libboost_wave-vc71-mt-gd-1_32.lib not > > > 'libboost_wave-vc7-mt-gd-1_32.lib'. I am not sure WHY the > > autolinking > > > is trying to hit the vc70 version of the lib since I am > > using VC71's IDE. > > > > > > (ANYONE HAVE ANY SUGGESTIONS?) > > > > I don't know; this looks like a possible misuse of the > > autolink system by Wave, or... maybe you need to > > force-rebuild your entire vc71 project? If you have any > > object files lying around that were build with vc70 you'll > > have the incorrect autolink references in them. > > Wave uses the autolink feature as described in the docs. No misuse here, > at > least I'm not aware of such. > > Regards Hartmut [Brian Braatz Writes:] I am using VC71 IDE, it is trying to link against the VC70 lib (yes? Look at the name) A bjam build of quickstart did not work for me. I deleted VC70 from my box, removed it from everything, and deleted all the Objs generated everywhere and I got things to compile. But I am not linking yet. I am flying today but will be working on this more over the weekend. I plan on re-building a fresh machine, installing the compiler and boost and going at it fresh (unless someone comes up with a better idea :) ) THANKS FOR YOUR HELP. Sincerely, "The Village Idiot" "Beware the fury of a patient man" - John Dryden

1 0

RE: [boost] Re: Customer Friendlier Boost Installation
by Brian Braatz 27 May '05

27 May '05

> -----Original Message----- > From: boost-bounces(a)lists.boost.org [mailto:boost-bounces@lists.boost.org] > On Behalf Of Rene Rivera > > Try this one... Gets you a good part of the way there.. > > http://boost-sandbox.sourceforge.net/boost-cvs-setup.exe > > [Brian Braatz Writes:] Allow me to also suggest an environment similar to Cgywin's. I have done this internal to our company and it has saved major problems in getting builds to actually work. If you find this interesting, I can attempt to explain it more.

1 0