- Boost - lists.boost.org

[property_tree review] Making the interface leaner, concretely...
by Ivan Vecerina 24 Apr '06

24 Apr '06

Hi, I wanted to give this a shot, as some others have agreed that the member-interface of ptree should be simplified - to being more like that of a container. I would plainly take out all the member functions labeled as "Ptree-specific operations" (see basic_ptree in the Synopsis in the main doc page: http://kaalus.atspace.com/ptree/doc/index.html ). A first step would be to convert them into a set of non-member functions (preferably in a separate header file). Another issue is that we have a multitude of functions because orthogonal features/concepts are being mixed together: - get, optional-get, default-get, put, and set (to be added) - own-data, child, and any-path access - for paths, custom or default separators - actual conversion of a value to/from a data string (or a ptree) - features such as first-class support for indexing an array also might be desirable Also what about: - stricter error checks: what happens if two children have the same name, making a path ambiguous? what about "hybrid" record/array tree nodes? - customization points: how do I specialize the conversion of a UDT to/from a ptree ? I would like to suggest as a starting point for this decomposition, to make the ptree library more flexible/adaptable to individual needs. How about the following building blocks: ? 1) value <-> ptree conversion templates void set( const Value& v, ptree& storage ); void get( Value& v, ptree const& storage ); //throws if conv. fails bool tryget( Value& v, ptree const& storage ) //ret. false if fails { try{ get(v,storage); return true; }catch(...){} return false; } The default implementation of these functions would use lexical_cast to store/retrieve a value to/from a ptree with only an 'own' value set. The template could of course be specialized for user-defined types. For example, I could implement these functions for a struct used as a member in various locations, and convert it into a ptree with a child for each struct data member. Which of get/tryget (throwing and non-throwing functions) is implemented in terms of the other is TBD. 2) Child access (direct child only) ptree const& getField( ptree const& tree, string childName ); // gets a field, throws in case of failure or ambiguity ptree const& trygetField( ptree const& tree, string childName ); // returns empty_ptree if the field is not found or ambiguous ptree& makeField( ptree& tree, string childName ); // accesses the field, creates it if needed ( like std::map::op[] ) + Do we want an "addField" function that always creates a new child even if one already exists? Or should we provide array-indexing functions ? 3) getoptional etc Shall be defined in terms of a sequence of calls to the previous functions. Exact set to be defined as discussed elsewhere. 4) path access TBD. I'll find it acceptable if existing get/put functions are preserved as non-member functions, but left outside of the class, and maybe preferably in a sub-namespace (ptree_stringpath, or whatever). Various proposals have been discussed in these review posts, and again could be implementable in terms of the above. Comments/opinions ? Ideas for using operator overloading? Ivan -- http://ivan.vecerina.com/contact/?subject=NG_POST <- email contact form

4 13

Re: [boost] [Boost Review] Property Tree Library
by Tom Brinkman 24 Apr '06

24 Apr '06

Sorry, But I cant follow you in the thread so I have to put this at the top level of the thread. I am not at home right now. > I understood it perfectly well. However, you also stated that the > parsers should come in at a later point. Thus my point still applies, if > only for limited time. However, that time is crucial. If there is a > Boost release (1.35) that includes the tree, but not the parser, people > might get used to that "useless property_tree thing" and not care to > look at its improved value in 1.36. Bad reputations tend to stick. Poor points. Getting this library thing ready by version 1.35 or any other version is my last concern. I want it to be correct. Its not acceptable that that this fairly large library will not get the full review that is necessary for a boost library. The scope of this library is too diverse for it that too occur in only a 10 day review period. It is very, very common for libraries to need a second or even a third review. I think that this library falls in that category of libraries that will need multiple reviews. > Also, the way I understood your post, your objection to the parsers was > that they are a completely separate thing from the tree and thus should > not be reviewed together. I find this objection simply wrong. The > parsers are a very important part of the property_tree library, and IMO > it's just its inappropriate name that makes them seem misplaced. As this seems to be opinion that you share with one or two others, please educate me. Why does the "property tree" require a parser. I just dont get it. How is the "property tree" container different from the vector and map containers, which almost certainly did not have parsers included along with them when they were added to the standard. What makes the "property tree" different from other containers that dont require parsers. I dont recall a single example of a container in C++ that had to have a parser. I suspect that those of you who want the parsers included with this library are just trying to sneek an XML parser through the review process without a full review. Please tell me that I'm wrong on this point. This library will eventually get accepted, I'm fairly sure of that, possibly even portions of the library will be approved this time around. What is your rush to get this one through. Lets take our time and get it right.

6 8

Overlap of property_trees with other boost libraries
by Ben Artin 24 Apr '06

24 Apr '06

The question was raised of possible overlap between the proposed property_tree library and the existing serialization and program_options libraries. My view on this matter: The primary purpose of the program_options library is to provide an input mechanism for simple (flat) data, specifically for the purpose of providing configuration options to programs that are configured via the command line or config files. The need for this library arises because those mechanisms for entering configuration options are common in command-line tools. The primary purpose of the serialization library is to provide a way to convert live C++ data structures into a form that can be transported to another program (or to the same program at a later time) and there reconstituted into a clone of the original data structure. The need for this library arises because the need for communication with other processes and storage of application data outside of the application process is common across many problem domains. The fact that the serialization library in its current form provides only conversions to byte streams is not important. This is just an implementation detail that arises from the fact that most common forms of interprocess communication and external storage use byte streams, and so conversion of C++ data to byte streams easily addresses both of those problems. Let us assume there there exists a data representation other than a byte stream that is suitable for external data representation or interprocess communication. Such a representation would be a reasonable archive format for the serialization library. A user of the serialization library could (at least in principle) take advantage of the strengths of this data representation. Regardless of the data representation used by the serialization library, one must treat the serialized format as opaque; there is a substantial amount of metadata embedded in it, and while the format can be made to be human-readable, it's predictably difficult to make it human-editable and even harder to make it human-writable. It it not possible to eliminate this metadata and retain other features provided by the serialization library (such as pointer tracking and class versioning). Now consider those design goals and constraints: the program_options library is an input method only; the serialization library is an input and output method, with features that prevent the data format from being human-editable. It is clear that in any case in which a human-editable input and output file format is needed, neither of these libraries will satisfy the requirement. As it turns out, the domain of persistent configuration data edited both by humans and by programs (such as configuration data of most GUI programs) is such a case, and that the need for persistent user-editable configuration data is not addressed by either program_options or serialization. Enter property_trees. This library attempts to fill this need (persistent user-editable data) and as a result overlaps with some functionality already provided by serialization and program_options. If all you need is data input, you could use either property_trees or program_options, as long as your data structures are simple enough to be representable by program_options. If you need input and output, you could use either serialization and property_trees, with different data format and code complexity tradeoffs. Here are some questions that I think need to be answered in order to decide whether this overlap is a good idea: 1) Is a property tree in-memory data structure a valuable abstraction, regardless of how it is serialized? I think the answer to this is yes. Representation of a hierarchy of key-value pairs is useful across the board, whether you are writing a compiler (symbol tables) or a GUI FTP client (cached per-directory data). 2) Is the serialization of such a data structure to formats less rich than those needed by the serialization library useful? I think the answer to this is yes. The obvious motivation for this is in serialization of configuration data to a user-editable form. 3) Could such serialization be done by the serialization library? I believe so. Provided that the serialization library has the notion of an archive that doesn't allow for class versioning or pointer tracking and may have limited ability to represent hierarchical data, many common configuration file formats (including Windows INI files, Mac OS X property lists, UNIX key<delimiter>value formats) can be produced by the serialization library. 4) Should such serialization be done by the serialization library? I believe so. The serialization library already has an interface for defining a way to convert C++ objects to an external representation, and I think it would be a good idea to maintain one interface for that purpose in boost. 5) Could a serialized form of property trees be used to input program options? Yes. So, here is one way that I see to resolve the property_trees conundrum: 1) A library providing an in-memory property tree abstraction should be submitted. 2) A library providing serialization of property trees to some common configuration file formats should be submitted. This could be part of the same library as in part 1. As part of this work, serialization library may need to be modified to allow for archive formats that cannot represent the full range of object and class tracking metadata. 3) Ideally, this should result in the ability to serialize arbitrary C++ objects directly to a configuration file format -- in-memory data shouldn't have to be in the explicit form of a property tree in order to be serializable to a configuration file. 4) Property tree file formats should be allowed as inputs to the program options library (and the existing program options configuration file format should be subsumed by the library from part 2 above). This would result in orthogonal components for each of the following responsibilities: 1) In-memory property tree representation 2) External property tree representation 3) Conversion of C++ objects, including in-memory property tree representation, to external property tree representation 4) Interpretation of property trees as command line program options 5) Parsing of command line program options from argv For what it's worth, Mac OS X has a native property tree API, and I have implemented a serialization archive format that produces such a property tree from C++ data structures using the serialization library. My understanding of the various abstractions and their interactions is based on that experience. Ben -- I changed my name: <http://periodic-kingdom.org/People/NameChange.php>

3 3

Re: [boost] Fwd: Program Options: string iterator not dereferencable
by Olaf van der Spek 24 Apr '06

24 Apr '06

On 4/24/06, Vladimir Prus <ghost(a)cs.msu.su> wrote: > On Monday 24 April 2006 12:19, you wrote: > > ---------- Forwarded message ---------- > > From: Olaf van der Spek <olafvdspek(a)gmail.com> > > Date: Apr 22, 2006 11:59 AM > > Subject: Program Options: string iterator not dereferencable > > To: boost(a)lists.boost.org > > > > > > Hi, > > > > I think there's a bug in the program options library. It appears to > > dereference an invalid string iterator. > > Hi Olaf, > I've just fixed this and now re-running all tests. I'll let you know when the > change is in. Thanks. There also seem to be 'typos' in some comments. Should I report those to you too? And what about: Is it possible to allow long options without -- prefix? The code doesn't seem to support it, but I think it's handy in cases where positional options aren't used.

1 0

Program Options: string iterator not dereferencable
by Olaf van der Spek 24 Apr '06

24 Apr '06

Hi, I think there's a bug in the program options library. It appears to dereference an invalid string iterator. libs\program_options\src\value_semantic.cpp: void validate(any& v, const vector<string>& xs, std::string*, int) { check_first_occurrence(v); string s(get_single_string(xs)); if (*s.begin() == '\'' && *s.rbegin() == '\'' || *s.begin() == '"' && *s.rbegin() == '"') v = any(s.substr(1, s.size()-2)); else v = any(s); } A !s.empty() check should be added.

2 1

[iostreams] Patch for BCB2006
by Nicola Musatti 23 Apr '06

23 Apr '06

Hallo, the enclosed patch brings support for the latest Borland compiler to the same level as bcc 5.6.4 . I tested it against bcc 5.8.2, bcc 5.6.4, VC++ 8.0 and g++ 3.4.2 . Cheers, Nicola Musatti ******Index: positioning.hpp =================================================================== RCS file: /cvsroot/boost/boost/boost/iostreams/positioning.hpp,v retrieving revision 1.8 diff -d -u -r1.8 positioning.hpp --- positioning.hpp 19 Nov 2005 19:56:13 -0000 1.8 +++ positioning.hpp 8 Apr 2006 14:09:33 -0000 @@ -46,7 +46,7 @@ return std::streampos(std::mbstate_t(), off); } -inline stream_offset fpos_t_to_offset(fpos_t pos) +inline stream_offset fpos_t_to_offset(std::fpos_t pos) { // Helper function. #if defined(_POSIX_) || (_INTEGRAL_MAX_BITS >= 64) return pos; Index: detail/bool_trait_def.hpp =================================================================== RCS file: /cvsroot/boost/boost/boost/iostreams/detail/bool_trait_def.hpp,v retrieving revision 1.6 diff -d -u -r1.6 bool_trait_def.hpp --- detail/bool_trait_def.hpp 12 Oct 2005 00:32:43 -0000 1.6 +++ detail/bool_trait_def.hpp 8 Apr 2006 14:09:57 -0000 @@ -20,7 +20,7 @@ // Description: Used to generate the traits classes is_istream, is_ostream, // etc. // -#if BOOST_WORKAROUND(__BORLANDC__, <= 0x564) +#if BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x582)) # define BOOST_IOSTREAMS_TRAIT_NAMESPACE(trait) #else # define BOOST_IOSTREAMS_TRAIT_NAMESPACE(trait) BOOST_PP_CAT(trait, _impl_):: Index: detail/config/codecvt.hpp =================================================================== RCS file: /cvsroot/boost/boost/boost/iostreams/detail/config/codecvt.hpp,v retrieving revision 1.3 diff -d -u -r1.3 codecvt.hpp --- detail/config/codecvt.hpp 1 Jun 2005 19:13:15 -0000 1.3 +++ detail/config/codecvt.hpp 8 Apr 2006 14:10:21 -0000 @@ -63,7 +63,7 @@ #if defined(__LIBCOMO__) using ::mbstate_t; -#elif defined(BOOST_DINKUMWARE_STDLIB) +#elif defined(BOOST_DINKUMWARE_STDLIB) && !defined(__BORLANDC__) using ::mbstate_t; #elif defined(__SGI_STL_PORT) #elif defined(BOOST_NO_STDC_NAMESPACE)

1 2

[io] Patch for BCB2006
by Nicola Musatti 23 Apr '06

23 Apr '06

Hallo, If I do not receive objections in the next few days I plan to apply the attached patch. Cheers, Nicola Musatti ******Index: ios_state.hpp =================================================================== RCS file: /cvsroot/boost/boost/boost/io/ios_state.hpp,v retrieving revision 1.4 diff -d -u -r1.4 ios_state.hpp --- ios_state.hpp 2 May 2005 06:24:20 -0000 1.4 +++ ios_state.hpp 18 Apr 2006 05:58:04 -0000 @@ -134,7 +134,11 @@ explicit basic_ios_exception_saver( state_type &s ) : s_save_( s ), a_save_( s.exceptions() ) {} +#if BOOST_WORKAROUND(__BORLANDC__, BOOST_TESTED_AT(0x582)) + basic_ios_exception_saver( state_type &s, aspect_type a ) +#else basic_ios_exception_saver( state_type &s, aspect_type const &a ) +#endif : s_save_( s ), a_save_( s.exceptions() ) { s.exceptions(a); } ~basic_ios_exception_saver()

1 0

boost::bimap [was Re: Do really property_tree and program_options overlap? ]
by JOAQUIN LOPEZ MU?Z 23 Apr '06

23 Apr '06

----- Mensaje original ----- De: Thorsten Ottosen <thorsten.ottosen(a)dezide.com> Fecha: Domingo, Abril 23, 2006 9:28 am Asunto: Re: [boost] Do really property_tree and program_options overlap? > Pavel Vozenilek wrote: > > "Thorsten Ottosen"wrote: > > > > > >>Well, I think we need both. Boost.MultiIndex is a great library > and can > >>do all kinds of wonderful things. But I would still like to see > >>a bidirectiobal map (boost::bimap) written as a wrapper around > it to > >>get an easy and specialized interface. > >> > > > > bimap is available in > > libs/multi-index/examples/bimap.cpp. > > Right, but the real value comes when somebody designs a nice > STL-like interface and write docs etc, at least that was my point. > Hello Thorsten, we have talked about this privately in the past, but I feel like bringing it to the list in the hope of getting the attention of potential contributors: There are some data structures buildable with B.MI which are regarded as particularly useful or common, like for instance the bidirectional map or bimap. A lean and mean implementation is provided in the aforementioned example, but certainly a much carefully crafted interface can be provided keeping B.MI as the implementation core: operator[], selection of 1-1/1-N/N-1/N-N variants, hashing/ordering, etc. I'm afraid I don't have the time to pursue this, as the current roadmap for core features of B.MI is taking all the spare time I can dedicate to the library. For this reason, I would love to see some volunteer jumping in who can develop this and other singular containers using B.MI (a cache container comes to mind) and then propose the results here either as a stand alone library of as part of B.MI --I'd prefer the former so as to keep the size of B.MI bounded. If there's such a volunteer I can provide her with some help/mentoring. I also wonder whether this is a task suitable to be proposed for Google Summer of Code. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

4 5

Re: [boost] [Boost Review] Property Tree Library
by Alexander 23 Apr '06

23 Apr '06

I am participating at project at which similar system was "invented". And so could provide some suggestions concerning the problem. We start our project from necessity to create windows registry like system for value storage at hierarchical order. Windows approach was quite adequate for our tasks but desire of compatibility makes it impossible to use it. After all win registry interface is archaic and inconvenient. It was our start point. We implemented interface which use node, leaf and attribute approach because of win registry influence. Now I could say that it's looks like mistake. This multilevel model makes interface too complicated without any benefits at reality. So, homogeneous interface of Property Tree Library is much more elegant and simple. I think that one tree implementation is not enough for all kinds of task. The simplest tree of course could be useful for small tasks. But it is not adequate for really big projects. So we create proxy which hides from us details of realization. It gives me opportunity to make unexpected solution for GUI. I don't like GUI programming and try to avoid it if only possible. So I create solution which allows users modify registry at safe manner. Methods of input, validators, access policy and so on prevents user from unsafe input. Safe behavior is based on attributes which stored at the same registry. But it's unsafe and inefficient to store data and visualization attributes at the same place. So proxy which mixed two different proxies for benefits of visualization was created. We decide that symbol '/' is forbidden for using as name. (If you wish to store filename or something which contains '/' you could store it as data not as key. Or if it so necessary to have '/' at name it is possible to use "//" mechanism. Point is not better at all because it's also could be at filename. 'Point looks strange as separator at hierarchical systems. Once more I prefer '/' after win registry semantic. Specifying separator at every function is ugly solution. ) For saving and loading we also use parser mechanism. But we have the basic class for it. It's my favorite trick to use vector of parsers for saving and loading. About name. Property Tree Library is not bad name I think, because being familiar with such approach I immediately understood what it's about. >What is your evaluation of the design? At least, the separator problem should be resolved adequately. >What is your evaluation of the implementation? >What is your evaluation of the documentation? Good enough >What is your evaluation of the potential usefulness of the library? I'm not quite sure but author do not positioning library as registry. I think this road should be leading to registry. At current form it is bonded to quite simple systems. >Did you try to use the library? With what compiler? no >Did you have any problems? no >How much effort did you put into your evaluation? It was only the comparison with my current solution. But it's taking a long time to. >Are you knowledgeable about the problem domain? Yes The Boost library at my opinion should resolve problem at full scale. It should provide foundation for full variety of different applications from small ini files to multithreaded and even dispersed registries. Alas, at current design library couldn't help me. But bear in mind that I am rather centered on my existing design and that library resolve more simple case I should vote for addition library to boost. Regards, Alexander Belyakov P.S. Well, I want make it clear to prevent misunderstanding that this library is not overlapped by boost.serialization at all but it will be overlapped by boost.registry if some one proposed it. It's clear enough for me that this library shouldn't be viewed as XML reader or any other reader. It's nice library with elegant and simple interface. Author defines property tree as a recursive data structure at another place he wrote it's a container. And it is really so. But REGESTRY is CONCEPTION not data structure. SERIALIZATION is also CONCEPTION. Author makes a lot of work. And he makes it almost ready to transform interface of container to something more interesting - to interface of conception. His library useful and I couldn't ask him to make a lot of work more. But if he wants he could make from it something more powerful. I could provide more information about my project if it's interesting.

1 0

Re: [boost] Do really property_tree and program_options overlap?
by JOAQUIN LOPEZ MU?Z 23 Apr '06

23 Apr '06

----- Mensaje original ----- De: Marcin Kalicinski <kalita(a)poczta.onet.pl> Fecha: Domingo, Abril 23, 2006 3:48 pm Asunto: Re: [boost] Do really property_tree and program_options overlap? > > I had now a look at the source code: If I understand correctly > it is > > possible to use the read/write functions (like read_xml, > read_ini, > > read_json etc.) with other containers? So there is no tight > coupling? > No, it is not normally possible. read/write functions expect a > recursive > container, because they need tree-like structure. I made no > attempt to > generalize the library, so that read/write functions can work with > general > recursive containers (btw. the whole idea of recursive container > is actually > quite fuzzy). What would it take to make the generalization? IMHO some concept of recursive container visitation/population along with an example of how your parsers can be made to work with data structures other than ptree would be a boon wrt to acceptance of the lib. Joaquín M López Muñoz Telefónica, Investigación y Desarrollo

1 0