From boris at codesynthesis.com Sun Jun 2 11:18:31 2013 From: boris at codesynthesis.com (Boris Kolpackov) Date: Sun Jun 2 11:17:13 2013 Subject: [odb-users] Building on Debian (testing/jessie) In-Reply-To: <51A7E1B3.9020705@adi-ware.ch> References: <51A7E1B3.9020705@adi-ware.ch> Message-ID: Hi Adrian, Adrian Imboden writes: > I found out about odb because of the very interesting C++Now presentation. I am glad you liked it! > When I want to build odb compiler, I get this output: > > # ./bootstrap > > [...] > > configure.ac:12: error: possibly undefined macro: m4_equote I am pretty sure you have checked out the code from the Git repository rather than using one of the packages. The problem is, the autotools (and VC++) build systems are auto-generated during the distribution preparation. So, for example, configure.ac in Git is a template that still has to be preprocessed. The easiest way to build ODB would be to use the distribution packages that we have prepared. Then you don't need to run bootstrap and can just go directly to configure and make. You can also check the INSTALL file for more information on this method. Alternatively, if you still want to use ODB from Git, then you can check the INSTALL-GIT file which describes how this can be done. Note that it requires quite a bit more initial setup compared to the first approach. > Is building on Debian supported or not? Yes, we make a complete Debian build from grounds up for every release. > I just upgraded my system from wheezy to Jessie because I suspected > that the required automake version is higher than my installed one > (Google suggested that to me). Actually, getting a combination of versions of autoconf, automake, and libtool that works reliably is quite tricky. The version of libtool that is available in Debian is very old and we use a newer version to prepare the release. That's why it is generally a good idea to avoid running bootstrap and rather use configure generated by the package maintainer. Boris From contact at rkade.fr Thu Jun 6 10:34:45 2013 From: contact at rkade.fr (rkadeFR) Date: Thu Jun 6 10:34:49 2013 Subject: [odb-users] Object creator with boost::posix_time::ptime member : exception from boost Message-ID: <51B09E05.2060701@rkade.fr> I have an existent database, and try to map the DB with my cpp objects for some queries. When I compiled, no errors. But when I run some queries on my objects, I have the exception "terminate called after throwing an instance of 'boost::exception_detail::clone_impl >'". I have some "0000-00-00 00:00:00" datetime in my DB (I can't change that). This exception is thrown exactly at the moment when I call a function on the iterator of my result. ex: iP->GetId() The documentation says that the object is created at this step of my program. (cf http://www.codesynthesis.com/products/odb/doc/manual.xhtml#4.4 ) So i put lots of log, and I can only see the private constructor called, and then... nothing except my exception. Can you explain me the way the iterator works ? And the work around I need to implement in order to avoid the exception ? Thank you, My object is something like : #include #include #include #pragma db object table("person") class Person { public: Person ( const std::string& name, const boost::posix_time::ptime& dateTime) : name(name), dateTime(dateTime) {} unsigned long GetId() const { return this->id; } const boost::posix_time::ptime GetDateTime() const { return this->dateTime; } const std::string GetName() const { return this->name; } private: Person() {} friend class odb::access; #pragma db column("id") id type("INT(10)") get(GetId) unsigned long id; #pragma db column("name") type("VARCHAR(25)") get(GetName) std::string name; #pragma db column("dateTime") type("DATETIME") get(GetDateTime) boost::posix_time::ptime tsAdded; } From boris at codesynthesis.com Thu Jun 6 11:38:48 2013 From: boris at codesynthesis.com (Boris Kolpackov) Date: Thu Jun 6 11:37:33 2013 Subject: [odb-users] Object creator with boost::posix_time::ptime member : exception from boost In-Reply-To: <51B09E05.2060701@rkade.fr> References: <51B09E05.2060701@rkade.fr> Message-ID: Hi, rkadeFR writes: > I have some "0000-00-00 00:00:00" datetime in my DB (I can't change that). > > So i put lots of log, and I can only see the private constructor > called, and then... nothing except my exception. > > Can you explain me the way the iterator works? The actual object loading happens when you dereference the iterator. What probably happens in your case is ODB is trying to load one of the objects with the "0000-00-00 00:00:00" value and since it is not valid for Boost ptime, an exception is thrown. > And the work around I need to implement in order to avoid the > exception ? The easiest would be to change your database to represent invalid datetimes (that's what I assume "0000-00-00 00:00:00" are used for) as NULL values. ODB will then map such NULL values to the special not_a_date_time value -- nice and clean. The alternative would be for us to add extra code to ODB in order to recognize all-zero datetime values and treat them as NULL. Boris From contact at rkade.fr Thu Jun 6 11:45:54 2013 From: contact at rkade.fr (rkadeFR) Date: Thu Jun 6 11:45:58 2013 Subject: [odb-users] Object creator with boost::posix_time::ptime member : exception from boost In-Reply-To: References: <51B09E05.2060701@rkade.fr> Message-ID: <51B0AEB2.9040502@rkade.fr> On 06/06/2013 17:38, Boris Kolpackov wrote: > Hi, > > rkadeFR writes: > >> I have some "0000-00-00 00:00:00" datetime in my DB (I can't change that). >> >> So i put lots of log, and I can only see the private constructor >> called, and then... nothing except my exception. >> >> Can you explain me the way the iterator works? > The actual object loading happens when you dereference the iterator. > What probably happens in your case is ODB is trying to load one of > the objects with the "0000-00-00 00:00:00" value and since it is > not valid for Boost ptime, an exception is thrown. Something I noticed: my member is a boost::posix_time::ptime object, but the exception is about a gregorian::date. I read the code of boost, and I assume that pTime includes a gregorian::date object. > >> And the work around I need to implement in order to avoid the >> exception ? > The easiest would be to change your database to represent invalid > datetimes (that's what I assume "0000-00-00 00:00:00" are used for) > as NULL values. ODB will then map such NULL values to the special > not_a_date_time value -- nice and clean. 1- I can't change the database 2- It's used as a negative infinite value :s > > The alternative would be for us to add extra code to ODB in order > to recognize all-zero datetime values and treat them as NULL. Can I change the constructor of the object? Or the copy? If I understand, ODB construct an empty object with the private cTor, and then fill the member with some copy? Can I change the behavior of this copy to handle the exception? > Boris From contact at rkade.fr Thu Jun 6 13:02:17 2013 From: contact at rkade.fr (rkadeFR) Date: Thu Jun 6 13:02:21 2013 Subject: [odb-users] Object creator with boost::posix_time::ptime member : exception from boost In-Reply-To: <51B0AEB2.9040502@rkade.fr> References: <51B09E05.2060701@rkade.fr> <51B0AEB2.9040502@rkade.fr> Message-ID: <51B0C099.2020709@rkade.fr> On 06/06/2013 17:45, rkadeFR wrote: > > On 06/06/2013 17:38, Boris Kolpackov wrote: >> Hi, >> >> rkadeFR writes: >> >>> I have some "0000-00-00 00:00:00" datetime in my DB (I can't change >>> that). >>> >>> So i put lots of log, and I can only see the private constructor >>> called, and then... nothing except my exception. >>> >>> Can you explain me the way the iterator works? >> The actual object loading happens when you dereference the iterator. >> What probably happens in your case is ODB is trying to load one of >> the objects with the "0000-00-00 00:00:00" value and since it is >> not valid for Boost ptime, an exception is thrown. > Something I noticed: my member is a boost::posix_time::ptime > object, but the exception is about a gregorian::date. I read the code > of boost, and I assume that pTime includes a gregorian::date object. >> >>> And the work around I need to implement in order to avoid the >>> exception ? >> The easiest would be to change your database to represent invalid >> datetimes (that's what I assume "0000-00-00 00:00:00" are used for) >> as NULL values. ODB will then map such NULL values to the special >> not_a_date_time value -- nice and clean. > 1- I can't change the database > 2- It's used as a negative infinite value :s >> >> The alternative would be for us to add extra code to ODB in order >> to recognize all-zero datetime values and treat them as NULL. > Can I change the constructor of the object? > Or the copy? I tried to change the behavior of the setter on another object. And I figure out that odb use the setter to fill the members of the objects through these setters. But as a setter, it takes already a const boost::posix_time::ptime& object, which is created before by ODB. I think the exception is thrown before by ODB than it reaches the setter method. void SetDateTime( const boost::posix_time::ptime& dateTime) { std::cout << "this is setter date time" << std::endl; this->dateTime = dateTime; } with the pragma... set(SetDateTime). > > If I understand, ODB construct an empty object with the private > cTor, and then fill the member with some copy? > Can I change the behavior of this copy to handle the exception? >> Boris > > From boris at codesynthesis.com Thu Jun 6 13:12:45 2013 From: boris at codesynthesis.com (Boris Kolpackov) Date: Thu Jun 6 13:11:28 2013 Subject: [odb-users] Object creator with boost::posix_time::ptime member : exception from boost In-Reply-To: <51B0AEB2.9040502@rkade.fr> References: <51B09E05.2060701@rkade.fr> <51B0AEB2.9040502@rkade.fr> Message-ID: Hi, rkadeFR writes: > 1- I can't change the database > 2- It's used as a negative infinite value :s Ok, so mapping it to neg_infin would be more appropriate. > If I understand, ODB construct an empty object with the private > cTor, and then fill the member with some copy? That's about right. The exception is thrown when ODB tries to initialize an instance of ptime from the loaded value, before assigning it to the data member in the object. > Can I change the behavior of this copy to handle the exception? Yes, ODB allows you to provide completely custom code that converts between the database value (in your case, that would be MYSQL_TIME struct) and C++ value (that would be ptime). There you can do pretty much any conversions you want. Also, if you are going this route, then you can also map DATETIME to something other than Boost ptime (e.g., your own date-time representation or struct tm). This guide has detailed information on how to do all this: http://www.codesynthesis.com/~boris/blog/2012/10/16/custom-cxx-to-database-type-mapping-in-odb/ You can also use the odb/boost/date-time/mysql/posix-time-traits.hxx file from the libodb-boost as a reference. There is also another option which doesn't require a custom value_traits specialization. It is a bit "hackish" (or cool; depending on how you look at it ;-)) and is better suited for once-off customizations that are only needed in specific classes. The overall idea is to use virtual data members and use the image type (MYSQL_TIME in our case) as its type. We will also need to provide our own accessors/modifiers that convert between our data member (or ptime type) and the image type. Here is an outline: #include // MYSQL_TIME class Person { ... void tsAddedConvert (const MYSQL_TIME& v) { using namespace boost::gregorian; using namespace boost::posix_time; if (v.year == 0 && v.month == 0 && v.day == 0 && v.hour == 0 && v.minute == 0 && v.second == 0) tsAdded = ptime (neg_infin); else tsAdded = ptime (date (v.year, v.month, v.day), time_duration (v.hour, v.minute, v.second)); } MYSQL_TIME tsAddedConvert () const { using namespace boost::gregorian; using namespace boost::posix_time; MYSQL_TIME v; v.neg = false; if (tsAdded.is_neg_infinity ()) { v.year = 0; v.month = 0; v.day = 0; v.hour = 0; v.minute = 0; v.second = 0; } else { const date& d (tsAdded.date ()); v.year = d.year (); v.month = d.month (); v.day = d.day (); const time_duration& t (tsAdded.time_of_day ()); v.hour = t.hours (); v.minute = t.minutes (); v.second = t.seconds (); } v.second_part = 0; return v; } #pragma db column("dateTime") type("DATETIME") #pragma db member(dateTime) virtual(MYSQL_TIME) access(tsAddedConvert) #pragma db transient boost::posix_time::ptime tsAdded; }; It is also possible to forward to the default ptime value_traits for the default cases (we would need to move the tsAddedConvert() functions to the .cxx file though). Boris From contact at rkade.fr Thu Jun 6 13:20:05 2013 From: contact at rkade.fr (rkadeFR) Date: Thu Jun 6 13:20:10 2013 Subject: [odb-users] Object creator with boost::posix_time::ptime member : exception from boost In-Reply-To: References: <51B09E05.2060701@rkade.fr> <51B0AEB2.9040502@rkade.fr> Message-ID: <51B0C4C5.6090902@rkade.fr> On 06/06/2013 19:12, Boris Kolpackov wrote: > Hi, > > rkadeFR writes: > >> 1- I can't change the database >> 2- It's used as a negative infinite value :s > Ok, so mapping it to neg_infin would be more appropriate. > > >> If I understand, ODB construct an empty object with the private >> cTor, and then fill the member with some copy? > That's about right. The exception is thrown when ODB tries to > initialize an instance of ptime from the loaded value, before > assigning it to the data member in the object. > > >> Can I change the behavior of this copy to handle the exception? > Yes, ODB allows you to provide completely custom code that converts > between the database value (in your case, that would be MYSQL_TIME > struct) and C++ value (that would be ptime). There you can do pretty > much any conversions you want. Also, if you are going this route, > then you can also map DATETIME to something other than Boost ptime > (e.g., your own date-time representation or struct tm). This guide > has detailed information on how to do all this: > > http://www.codesynthesis.com/~boris/blog/2012/10/16/custom-cxx-to-database-type-mapping-in-odb/ > > You can also use the odb/boost/date-time/mysql/posix-time-traits.hxx > file from the libodb-boost as a reference. > > There is also another option which doesn't require a custom value_traits > specialization. It is a bit "hackish" (or cool; depending on how you look > at it ;-)) and is better suited for once-off customizations that are only > needed in specific classes. The overall idea is to use virtual data members > and use the image type (MYSQL_TIME in our case) as its type. We will also > need to provide our own accessors/modifiers that convert between our > data member (or ptime type) and the image type. Here is an outline: > > #include // MYSQL_TIME > > class Person > { > ... > > void tsAddedConvert (const MYSQL_TIME& v) > { > using namespace boost::gregorian; > using namespace boost::posix_time; > > if (v.year == 0 && > v.month == 0 && > v.day == 0 && > v.hour == 0 && > v.minute == 0 && > v.second == 0) > tsAdded = ptime (neg_infin); > else > tsAdded = ptime (date (v.year, v.month, v.day), > time_duration (v.hour, v.minute, v.second)); > } > > MYSQL_TIME tsAddedConvert () const > { > using namespace boost::gregorian; > using namespace boost::posix_time; > > MYSQL_TIME v; > v.neg = false; > > if (tsAdded.is_neg_infinity ()) > { > v.year = 0; > v.month = 0; > v.day = 0; > > v.hour = 0; > v.minute = 0; > v.second = 0; > } > else > { > const date& d (tsAdded.date ()); > v.year = d.year (); > v.month = d.month (); > v.day = d.day (); > > const time_duration& t (tsAdded.time_of_day ()); > v.hour = t.hours (); > v.minute = t.minutes (); > v.second = t.seconds (); > } > > v.second_part = 0; > return v; > } > > #pragma db column("dateTime") type("DATETIME") > #pragma db member(dateTime) virtual(MYSQL_TIME) access(tsAddedConvert) > > #pragma db transient > boost::posix_time::ptime tsAdded; > }; > > It is also possible to forward to the default ptime value_traits for > the default cases (we would need to move the tsAddedConvert() functions > to the .cxx file though). > > Boris Ok great, I'm gonna try to implement my own type :) From PStath at Axxcelera.com Thu Jun 6 14:59:38 2013 From: PStath at Axxcelera.com (Stath, Paul) Date: Thu Jun 6 14:59:56 2013 Subject: [odb-users] Enable RSS feed for odb-users? Message-ID: <5B5D636C5F0AA747B37D0EFC72B563C404B12251@RIC-MS02.MWSG.int> Boris -- Would it be possible to get an RSS feed for odb-users, similar to the RSS feed "ODB News Feed"? (I'm making the perhaps incorrect assumption that "ODB News Feed" is an RSS of the odb-announcements list.) I'm in the process of attempting to lighten my in-box by moving my mailing list subscriptions to RSS feeds where available. -- Paul From tenchu.tarik at hotmail.fr Wed Jun 5 12:12:53 2013 From: tenchu.tarik at hotmail.fr (Tarik BENZ) Date: Thu Jun 6 15:48:07 2013 Subject: [odb-users] Join Between Tables Message-ID: Hello, I Have three tables that I mapped, and I would like to know how to do a join between them. Do I necessarily need to create a view? Best Wishes, Tarik From boris at codesynthesis.com Thu Jun 6 16:01:06 2013 From: boris at codesynthesis.com (Boris Kolpackov) Date: Thu Jun 6 16:00:01 2013 Subject: [odb-users] Join Between Tables In-Reply-To: References: Message-ID: Hi Tarik, Tarik BENZ writes: > I Have three tables that I mapped, and I would like to know > how to do a join between them. Do I necessarily need to create a view? ODB automatically creates a JOIN in two cases (well, technically, there is a third case: polymorphic objects, but that is not generally visible to the end user): 1. When there is a relationship between objects (i.e., one points to the other). In this case, ODB "JOIN's in" the pointed-to objects in queries so that you can refer to data members from these objects. For example: db.query (query::employer->name == "Example, Inc"); Note that this JOIN'ing of related objects is only done one level deep. 2. In views that are based on more than one object (or database table). There is a lot of flexibility in views and you can JOIN objects and/or database tables using relationships and/or custom conditions. So, unless the JOIN that you are looking for naturally follows from the relationship between your objects, the views are probably your best choice. Boris From boris at codesynthesis.com Fri Jun 7 03:09:12 2013 From: boris at codesynthesis.com (Boris Kolpackov) Date: Fri Jun 7 03:07:56 2013 Subject: [odb-users] Enable RSS feed for odb-users? In-Reply-To: <5B5D636C5F0AA747B37D0EFC72B563C404B12251@RIC-MS02.MWSG.int> References: <5B5D636C5F0AA747B37D0EFC72B563C404B12251@RIC-MS02.MWSG.int> Message-ID: Hi Paul, Stath, Paul writes: > I'm in the process of attempting to lighten my in-box by moving my > mailing list subscriptions to RSS feeds where available. Have you looked at the digest option? In essence, instead of receiving every post individually (and in real time), you can configure your subscription to send you a single email with all the accumulated posts once in a while. You can enable digest for your subscription here: http://www.codesynthesis.com/mailman/options/odb-users > Would it be possible to get an RSS feed for odb-users, similar to the > RSS feed "ODB News Feed"? I took a look and there doesn't seem to be an easy way to add this support in mailman (it is on their TODO list). Also, my problem with an RSS feed of a mailing list is that there is no easy way to reply to a post. Before you could have written a quick reply to a question you know the answer to because it is so easy. With RSS you probably won't bother. If you still want an RSS feed, then I think the best option is to add odb-users to gmane.org, which provides all kinds of RSS feeds for the mailing lists it tracks. > (I'm making the perhaps incorrect assumption that "ODB News Feed" is > an RSS of the odb-announcements list.) Yes, the two provide the same information and the RSS feed is hand- maintained. This works well because odb-announcements is a read-only mailing list. Boris From contact at rkade.fr Fri Jun 7 07:10:46 2013 From: contact at rkade.fr (rkadeFR) Date: Fri Jun 7 07:10:52 2013 Subject: [odb-users] Object creator with boost::posix_time::ptime member : exception from boost In-Reply-To: <51B0C4C5.6090902@rkade.fr> References: <51B09E05.2060701@rkade.fr> <51B0AEB2.9040502@rkade.fr> <51B0C4C5.6090902@rkade.fr> Message-ID: <51B1BFB6.5050307@rkade.fr> On 06/06/2013 19:20, rkadeFR wrote: > > On 06/06/2013 19:12, Boris Kolpackov wrote: >> Hi, >> >> rkadeFR writes: >> >>> 1- I can't change the database >>> 2- It's used as a negative infinite value :s >> Ok, so mapping it to neg_infin would be more appropriate. >> >> >>> If I understand, ODB construct an empty object with the private >>> cTor, and then fill the member with some copy? >> That's about right. The exception is thrown when ODB tries to >> initialize an instance of ptime from the loaded value, before >> assigning it to the data member in the object. >> >> >>> Can I change the behavior of this copy to handle the exception? >> Yes, ODB allows you to provide completely custom code that converts >> between the database value (in your case, that would be MYSQL_TIME >> struct) and C++ value (that would be ptime). There you can do pretty >> much any conversions you want. Also, if you are going this route, >> then you can also map DATETIME to something other than Boost ptime >> (e.g., your own date-time representation or struct tm). This guide >> has detailed information on how to do all this: >> >> http://www.codesynthesis.com/~boris/blog/2012/10/16/custom-cxx-to-database-type-mapping-in-odb/ >> >> >> You can also use the odb/boost/date-time/mysql/posix-time-traits.hxx >> file from the libodb-boost as a reference. >> >> There is also another option which doesn't require a custom value_traits >> specialization. It is a bit "hackish" (or cool; depending on how you >> look >> at it ;-)) and is better suited for once-off customizations that are >> only >> needed in specific classes. The overall idea is to use virtual data >> members >> and use the image type (MYSQL_TIME in our case) as its type. We will >> also >> need to provide our own accessors/modifiers that convert between our >> data member (or ptime type) and the image type. Here is an outline: >> >> #include // MYSQL_TIME >> >> class Person >> { >> ... >> >> void tsAddedConvert (const MYSQL_TIME& v) >> { >> using namespace boost::gregorian; >> using namespace boost::posix_time; >> >> if (v.year == 0 && >> v.month == 0 && >> v.day == 0 && >> v.hour == 0 && >> v.minute == 0 && >> v.second == 0) >> tsAdded = ptime (neg_infin); >> else >> tsAdded = ptime (date (v.year, v.month, v.day), >> time_duration (v.hour, v.minute, v.second)); >> } >> >> MYSQL_TIME tsAddedConvert () const >> { >> using namespace boost::gregorian; >> using namespace boost::posix_time; >> >> MYSQL_TIME v; >> v.neg = false; >> >> if (tsAdded.is_neg_infinity ()) >> { >> v.year = 0; >> v.month = 0; >> v.day = 0; >> >> v.hour = 0; >> v.minute = 0; >> v.second = 0; >> } >> else >> { >> const date& d (tsAdded.date ()); >> v.year = d.year (); >> v.month = d.month (); >> v.day = d.day (); >> >> const time_duration& t (tsAdded.time_of_day ()); >> v.hour = t.hours (); >> v.minute = t.minutes (); >> v.second = t.seconds (); >> } >> >> v.second_part = 0; >> return v; >> } >> >> #pragma db column("dateTime") type("DATETIME") >> #pragma db member(dateTime) virtual(MYSQL_TIME) >> access(tsAddedConvert) >> >> #pragma db transient >> boost::posix_time::ptime tsAdded; >> }; >> >> It is also possible to forward to the default ptime value_traits for >> the default cases (we would need to move the tsAddedConvert() functions >> to the .cxx file though). >> >> Boris > Ok great, > > I'm gonna try to implement my own type :) > > Ok ! Done for what I wanted :) Thank you so much :) From PStath at Axxcelera.com Fri Jun 7 12:03:20 2013 From: PStath at Axxcelera.com (Stath, Paul) Date: Fri Jun 7 15:04:31 2013 Subject: [odb-users] Enable RSS feed for odb-users? In-Reply-To: References: <5B5D636C5F0AA747B37D0EFC72B563C404B12251@RIC-MS02.MWSG.int>, Message-ID: <5B5D636C5F0AA747B37D0EFC72B563C404B12CD9@RIC-MS02.MWSG.int> Boris -- > Have you looked at the digest option? In essence, instead of receiving I'm already using digest mode. I don't have a problem w/ the velocity of odb-users, but do have a problem with a couple of the other mailing lists that I subscribe to. (Digest doesn't make it much better either.) So I'm attempting to get all of my mailing list subs as RSS feeds. > I took a look and there doesn't seem to be an easy way to add this > support in mailman (it is on their TODO list). I too looked to see if GNU Mailman provided RSS feeds, but no one has provided one. (I'm actually amazed, as I think this would be extremely useful. And would be a logical function for the archiver to perform.) > Also, my problem with an RSS feed of a mailing list is that there is > no easy way to reply to a post. Before you could have written a quick > reply to a question you know the answer to because it is so easy. With > RSS you probably won't bother. This is a good point. It is not as easy as a simple email reply. Of course since I use digest mode, you have almost always responded before I even see the question. ;) > If you still want an RSS feed, then I think the best option is to add > odb-users to gmane.org, which provides all kinds of RSS feeds for the > mailing lists it tracks. This is how I have handled the other mailing lists I subscribe to. I had not realized that I could simply request a new list to be tracked. That will work for me. Thanks. -- Paul From jneuhart at tlirr.com Fri Jun 7 16:36:31 2013 From: jneuhart at tlirr.com (Jordan J. Neuhart) Date: Fri Jun 7 16:36:38 2013 Subject: [odb-users] Polymorphic id name not carried through to dervied table Message-ID: <681FA42AB546F84C8CFDFCAA6D89A62802E727A9@husker.tlirr.local> Greetings, I have a polymorphic class hierarchy and when I check the generated SQL files, I see that the column name that I specified for the id column of the base class is not the same as the name of the id column for the derived class. I have attached the generated files so that you can see this. I am using the Qt profile and the SQLite odb library on Windows 7. I am using the latest version of the odb compiler. Thanks, Jordan Neuhart Database Administrator Sofware Developer T-L Irrigation Co. P.O. Box 1047 151 East Hwy 6 & AB Road Hastings, NE 68902-1047 office: 402-462-4128 ext. 264 800-330-4264 ext. 264 cell: 402-217-1377 email: jjn@tlirr.com website: http://www.tlirr.com From boris at codesynthesis.com Sat Jun 8 09:26:51 2013 From: boris at codesynthesis.com (Boris Kolpackov) Date: Sat Jun 8 09:25:36 2013 Subject: [odb-users] Polymorphic id name not carried through to dervied table In-Reply-To: <681FA42AB546F84C8CFDFCAA6D89A62802E727A9@husker.tlirr.local> References: <681FA42AB546F84C8CFDFCAA6D89A62802E727A9@husker.tlirr.local> Message-ID: Hi Jordan, Jordan J. Neuhart writes: > I have a polymorphic class hierarchy and when I check the generated SQL > files, I see that the column name that I specified for the id column of > the base class is not the same as the name of the id column for the > derived class. Yes, that was an oversight. We should propagate the custom column name to the derived classes. I've fixed this and you can get the ODB compiler with this fix here: http://www.codesynthesis.com/~boris/tmp/odb/odb-2.2.2-i686-windows.zip Boris From vwygos at gmail.com Sun Jun 9 13:32:44 2013 From: vwygos at gmail.com (Piotr Wygocki) Date: Sun Jun 9 14:57:06 2013 Subject: [odb-users] ODB Message-ID: Hi' I've been to your presentation on c++now. I like ODB very much but there is one think that bothers me. The think is that you cannot run which translates directly to this kind of query: "UPDATE table SET id = id + 1 WHERE id > 10". or "DELETE FROM table WHERE id > 10". In my previous company I used dbs a lot and lack of this kind of queries would be disqualifying for a framework. Good luck with future developement! Regards, Piotr From boris at codesynthesis.com Sun Jun 9 15:02:56 2013 From: boris at codesynthesis.com (Boris Kolpackov) Date: Sun Jun 9 15:01:40 2013 Subject: [odb-users] ODB In-Reply-To: References: Message-ID: Hi Piotr, Piotr Wygocki writes: > I like ODB very much but there is one think that bothers me. The > think is that you cannot run which translates directly to this kind > of query: > > "UPDATE table SET id = id + 1 WHERE id > 10". Yes, we call these "mass update" statements (since they affect multiple objects). There are plans to support mass updates (and mass, inserts for databases that support them) in the near future. > "DELETE FROM table WHERE id > 10". This one is actually already supported: db.erase_query