- Boost - lists.boost.org

[1.41.0] Cleaning up inspection errors
by Beman Dawes 09 Oct '09

09 Oct '09

The Inspection Report is showing 1292 problems reported. I'd like to drive that number way lower for the 1.41.0 release. The "Worst Offenders" list is: test (254) regression (145) gil (140) build (116) proto (75) See http://boost.cowic.de/rc/inspect-trunk.html Developers, please fix any problems in your library. Thanks, --Beman

2 2

[program-options] Ticket #1132, multitoken() and zero_token()
by Sascha Ochsenknecht 09 Oct '09

09 Oct '09

Hello, I would like to have an option which allows multi token but not multiple occurrences of the option or at least a way to distinguish the occurrences: E.g. something like: program.exe --option value1 value2 I tried the following: 1) desc.add_options()("option", value<string>()->multitoken(), "help"); With the above command line I get exception: multiple values not allowed 2) desc.add_options()("option", value<vector<string> >(), "help"); Command line is parsed without error but only value1 is stored in the vector -> ERROR, should report 3) desc.add_options()("option", value<vector<string> >()->multitoken(), "help"); Command line is parsed without error both values are stored in the vector, but program.exe --option value1 --option value2 and program.exe --option value1 value2 leads to the same vector contents. With the current version it isn't possible to distinguish to occurrences of a certain option and it is not possible to implement an option which has the multitoken attribute but isn't allowed to occur several times. I'm willing to work on a patch for this issue, but first I want to discuss how to solve this, i.e. how the library should behave. The problem is possibly that specifying a list or vector implies that an option can occur multiple times. I think maybe following solution could work: 1) desc.add_options()("option", value<vector<string> >(), "help"); multiple occurrences of the option are allowed and pushed to the vector but only one value per option occurrence (like it is implemented now) 2) desc.add_options()("option", value<vector<string> >()->multitoken(), "help"); multiple occurrences of the option are NOT allowed, the multiple tokens/values are pushed to the vector 3) desc.add_options()("option", value<vector<vector<string> > >()->multitoken(), "help"); multiple occurrences of the option are allowed, the multiple tokens/values are pushed to the inner vector E.g. program.exe --option value1 value2 --option value3 --option value4 value5 This would result in the following storage sheme: { { value1, value2 } { value3 } { value4 value5 } } I'm not sure if a could become a friend of this implementation ... but it would work at least. Another way would be to join the multitokens into a single string: { "value1 value2" "value3" "value4 value5" } .. but this I don't like :-(. What do you think? Thanks & regards, Sascha PS: Is somebody working on ticket #1132 ??

1 0

[rdb] 0.11
by Jean-Louis Leroy 08 Oct '09

08 Oct '09

In this drop : * Everything related to SQL composition has been moved to namespace boost::rdb::odbc * Update syntax has changed, instead of .set().set() chains it is now closer to SQL syntax (an deasier to implement): update(p).set(p.id = 1, p.first_name = "Homer") * Qualified table names: person::qualified p; partner::qualified l; select(p.id).from(p, l).where(p.id == l.wife) gives: select person.id from person, partner where person.id = partner.wife * in predicate: select(p.id).from(p, l).where(p.id.in(select(l.husband).from(l))) select(p.id).from(p).where(p.id.in(p.age + 1, 2)) * NULL: insert_into(p)(p.id, p.age).values(1, null) update(p).set(p.age = null).where(p.id == 1) select(p.id).from(p).where(p.age == null) Whoa! This gives: select id from person where age is null I'm only 80% sure about this one. Like everybody I suppose I have tried one day to compare with NULL, only to realize that the result is always false. After a while, ahem...So I have implemented operator == to do the right thing. It could have been: select(p.id).from(p).where(p.age.is_null()) Does anyone has an example where it makes sense to truly compare (using =) with NULL ? As for returning the NULL indicator when reading results, I chose not to use Boost.Optional (although I will probably add support for optional as well) because: 1. it would cost one bool (most likely the size of an integer) per column 2. IMHO the automatic conversion to bool is a bad idea 3. the value inside an optional<> cannot be addressed Instead I have complemented a fusion::vector with a std::bitset. The tutorial shows how to use this. I am hesitating a bit about pushing fusion::vector inside the implementation (i.e. nullable<int, char...> instead of nullable< fusion::vector<int, char> >). * Result sets: execute(Select) now returns a ResultSet object that allows retrieval row by row: also see the tutorial here : http://yorel.be/boost/libs/rdb/doc/html/rdb/tutorial.html#retrieve_data There is also a "Jumbo select" at the bottom of the tutorial that recaps everything that is supported for the moment. J-L

5 10

[threads] wait() and signal() ?
by David M. Cotter 07 Oct '09

07 Oct '09

on the mac the notion of a semaphore is at a slightly higher level than a condition_variable. but it seems in boost::thread we must use the actual condition_variable. on the mac, with a semaphore, the value can be pre-set to signal, so that the next wait() is passed right thru, since the signal was sent ahead of time. eg: you could do this: Semaphore sem; // newly initialized semaphore, brand new, never waited upon sem.signal(); // you can pre-set the signal! // later in the code sem.wait(); // no wait, since signal was already sent // and later still sem.wait(); // now we wait in fact, actually, the way it actually happens on the mac is that a newly initialized semaphore comes IN WITH the semaphore pre-signaled. so you actually have to call wait() immediately to eat that signal. i've got lots of code that does this: Semaphore sem; sem.wait(); // eat the first signal i thought that was normal. trouble is,i've got lots of code that does NOT eat that first signal, depending on the first wait call to be passed thru immediately. this is the code that now breaks. in converting to boost::thread, there doesn't seem to be a way to pre- set the signal. or to reset the state of a semaphore from an unknown state to a known state, eg this code: sem.signal(); sem.signal(); sem.wait(); that guarantees that the next wait will actually wait, regardless of the state of the sem before hand. how can i do this with boost::thread ?

6 7

[rdb] Repository, contributing, copyright, etc
by Jean-Louis Leroy 07 Oct '09

07 Oct '09

* Respository I have set up a Mercurial repository for rdb. You can get a copy with this command: hg clone https://boost-rdb.googlecode.com/hg/ boost-rdb You can view the source without installing Mercurial at that URL. * Contributing For those who consider contributing: I would like to keep clear copyright on this work. I propose the following arrangement: if you send me a patch that touches any of my copyrighted files, and I decide to use it, then you agree to transfer the copyright on the entire patch content to me. In return, I guarantee: 1/ the files touched by the patch will remain available under the Boost license 2/ you will get credit for it in the "Acknowledgements" section of the doc. If you create a set of files that is independant from mine and wish to put it in the distributions or in the repository, then you retain the copyright on them and set the rules you want. What I have in mind here is a native binding to a database. I hope that by next year we will have native bindings for all the major databases and obviously I am not going to write them all by myself. * Please help I am looking for the sql 2003 syntax expressed as a graph, preferably available online. Books suggestions are welcome too. * Extensibility For a small touch of technical stuff in this mail: I am trying hard to make it easy to support the various dialects. Here is a proof of concept (add MySQL LIMIT clause): https://boost-rdb.googlecode.com/hg/libs/rdb/doc/examples.cpp J-L

1 0

[utility] auto_buffer v2
by Thorsten Ottosen 07 Oct '09

07 Oct '09

Dear All, Hereby an updated class. Thanks to all who contributed with comments to the first try. I have made the following changes: 1. push_back() overloads are now growing 2. added swap() (boy, was this nasty to implement). I assume the allocator can be wapped without throwing. How do other people copy with this problem? 3. added copy-constructor + copy-assignment 4. made the min capacity equal to N, the number of elements in the stack buffer 5. the size of the internal buffer is now controlled via two policies: template< unsigned N > struct store_n_objects; template< unsigned N > struct store_n_bytes; the default is store_n_objects<256>. The reason is that we don't want the capacity to change just because we compile on a different platform. 6. added comparison operators Comments most welcome! best regards -Thorsten

4 7

[asio] confused by multiple socket flavors
by Roland Schwarz 07 Oct '09

07 Oct '09

There is a stream_socket_service and a ip::tcp::socket class, with almost the same member functions. I do understand that the first "is_a" service while the second "has_a" service. Is there any rationale beyond just having both? When would I want to use which one? Do I have overlooked something in the documentaion? Any hints appreciated. Roland -- _________________________________________ _ _ | Roland Schwarz aka. speedsnail |_)(_ | sip:speedsnail@ekiga.net | \__) | mailto:roland.schwarz@chello.at ________| http://www.blackspace.at

1 0

custom async devices
by Roland Schwarz 07 Oct '09

07 Oct '09

Recently I had a question about how I could use asio for performing asynchroneous file IO. In the meantime I found that there are already some classes for doing async_read_at IO on handles, but only for windows. Since I need to address multiple platforms I am wondering how I would go about that. I am thinking of an emulation layer in user-space based on threads. I guess I could set up a thread-pool that will accept async call ins. The thread pool would know about an io_service and when a task, e.g. a file IO has completed, use the post function on the io_service to invoke the completion handler. Is this a reasonable design? Any pros/cons? Does anyone know of an example that I could use as a starter? Thank you, Roland -- _________________________________________ _ _ | Roland Schwarz aka. speedsnail |_)(_ | sip:speedsnail@ekiga.net | \__) | mailto:roland.schwarz@chello.at ________| http://www.blackspace.at

1 0

how to mark test as "expected fail" for some toolsets
by Gunter Winkler 07 Oct '09

07 Oct '09

Hello, I'd like to mark the begin_end test of uBLAS as compile-fail for MSVC 7.1 ([1]). However I was not able to figure out how this works. Does anyone have some insights (or maybe a patch) to do this? [1] http://www.boost.org/development/tests/trunk/developer/output/daw-msvc71-bo… mfg Gunter

3 3

Bjam outside Boost tree
by Jean-Louis Leroy 06 Oct '09

06 Oct '09

I am working on a lib that is laid out in typical boost fashion: a directory with boost/rdb/includes.hpp, libs/rdb/src/stuff.cpp. When I put these two directories right inside a Boost tree that builds properly with bjam, I can build my libs as well (via chdir to libs/rdb/test and run bjam). I managed to get that much working. Now I would like to build outside the boost tree. I have found this message:http://lists.boost.org/boost-build/2002/11/2221.php but the link it contains is broken. I have tried to locate the information in the doc for half an hour now, without success. I tried to set BOOST_BUILD_PATH AND BOOST_ROOT to point to the boost tree. And copied boost-build.jam to the root of my dir. Can someone help me with this please? TIA, J-L

3 14