Re: [sqlalchemy] Re: SQLAlchemy 0.9.1 released
I found a problem in 0.9.1 version: in 0.8.x : print (Blog.c.id==5) None blog.id = :id_1 But in 0.9.1: print (Blog.c.id==5) None blog.id = :id_1 AND NULL So I don't know if it's a bug? On Tue, Jan 7, 2014 at 2:25 AM, Jonathan Vanasco jonat...@findmeon.comwrote: automap sounds neat! thanks! -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out. -- I like python! UliPad The Python Editor: http://code.google.com/p/ulipad/ UliWeb simple web framework: https://github.com/limodou/uliweb My Blog: http://my.oschina.net/limodou -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out.
[sqlalchemy] Cubes OLAP 1.0alpha with new SQL backend features
Hi, First, happy new year to you all and thanks for the great work on SQLAlchemy. I just would like to share a bit of news. Cubes – Python Lightweight OLAP framework is approaching 1.0 with quite a few new things. Cubes is using SQLAlchemy for the SQL ROLAP backend. The notable SQL related changes are: * multiple data stores[1] (databases/connections) per server session * outer joins for dimensions [2] – allows you to include dimension members in the result even if there are no related facts. * named model-wide join specification – you can specify a join (dimension table, column, type, ...) once in the model to be easily reused in any cube just by adding cube-specific attribute (just fact key(s)) The outer joins feature required a major rewrite of the SQL star schema browser. The nature of SQLAlchemy helped a lot. Some notable non-SQL related changes: * distinction between measures and aggregates * new backends: Mongo, Mixpanel, Google Analytics, Slicer More extensive list of 1.0 changes can be found at [3] Note: There are many necessary backward incompatible changes. Please refer to the documentation for more information. Links: [1] http://cubes.databrewery.org/dev/doc/workspace.html [2] http://cubes.databrewery.org/dev/doc/backends/sql.html#sql-outer-joins [3] http://cubes.databrewery.org/dev/doc/releases/1.0.html Comments, bug reports and questions are welcome. The Cubes Goggle Group is: https://groups.google.com/forum/#!forum/cubes-discuss Cheers, Stefan -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sqlalchemy] SQLAlchemy 0.9.1 released
that’s this: http://docs.sqlalchemy.org/en/rel_0_9/changelog/migration_09.html#improved-rendering-of-boolean-constants-null-constants-conjunctions On Jan 7, 2014, at 4:13 AM, limodou limo...@gmail.com wrote: I found a problem in 0.9.1 version: in 0.8.x : print (Blog.c.id==5) None blog.id = :id_1 But in 0.9.1: print (Blog.c.id==5) None blog.id = :id_1 AND NULL So I don't know if it's a bug? On Tue, Jan 7, 2014 at 2:25 AM, Jonathan Vanasco jonat...@findmeon.com wrote: automap sounds neat! thanks! -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out. -- I like python! UliPad The Python Editor: http://code.google.com/p/ulipad/ UliWeb simple web framework: https://github.com/limodou/uliweb My Blog: http://my.oschina.net/limodou -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out. signature.asc Description: Message signed with OpenPGP using GPGMail
[sqlalchemy] matches_any: an extension to the Query object, and a HSTORE attribute access property
Hi folks, SQLAlchemy is at the heart of what we do at work (UK based hedge fund); so first of all a big thank you; I am not quite sure where we'd be without you. We would like to give back to the community as much as we can (I tried to get some of the developers on the company xmas present list this year but was too late.. cross your fingers for next year). We have extended SQLAlchemy in a few places, though it is quite intermingled with our domain specific stuff I keep an eye out for little snippets that might be useful to others. So here's a trivial one; take it or leave it (and feel free to think of a better name). Knowing my luck it already exists; though I have looked hard through the docs! class QueryEnhanced(Query): ''' Add a few extra bells and whistles to the standard Query object ''' def matches_any(self): ''' Returns true if your query would return 1 or more rows; false otherwise. The following two statements ask the same question; but matches_any is _much_ quicker on large tables: my_query.matches_any() my_query.count() != 0 ''' return self.session.scalar(select([self.exists()])) The other bit of technology we have that could be unpicked without _too_ much trouble is a sort of reverse CompositeProperty; many attributes of different types, including collections, out of one HSTORE column (with a sort of side-loaded instrumentation for mutation tracking that I think could have been done in a more idiosyncratic way). Paraphrasing a bit but you can do things like: class Animal(Base): data = Column(MutableDict.as_mutable(HSTORE)) colour = HsProperty(data, String) legs = HsProperty(data, Integer) discovered = HsProperty(data, Date) fun_facts = HsProperty(data, JSONEncoded(list)) 'colour', 'legs', 'discovered', and 'fun_facts' end up as keys in the HSTORE and the values are strings, integers, dates and lists on the python side but stored as strings in the HSTORE such a way that they can be CAST-ed in a server query [where possible]: session().query(Animal).filter(Animal.legs 2) and get a query like SELECT ... FROM animal WHERE CAST(animal.data - legs AS INTEGER) 2 You can also put an arbitrary JSONEncodable object in there too. Collections get converted to Mutable counterparts for change-tracking. In many ways it is similar to ColumnProperty except that - the properties are writable (and when written only trigger the relevant bits of the hstore to be updated). Also on object instances the values in HsProperties are fetched as part of the query; we lazily de-serialise them directly from the hstore dictionary. Before spend a couple of days removing our corporate clutter from that, getting permission to license it etc.. and posting either as a patch or extension I thought I would see if there is any interest (or if someone has already done it better?). It's implemented as a custom metaclass right now, but I think I might be able to do it fully with events. Code aside, if you can think of ways in which we as a company could support SQLAlchemy (bear in mind I am not in charge of the purse strings, but I can make a pitch on your behalf; we are still awaiting the fruits of our donation to the PyPy http://morepypy.blogspot.co.uk/2012/01/py3k-and-numpy-first-stage-thanks-to.htmlguys :).** Then do let me know. I don't check this email account all that regularly but my work address is my firstname.lastname at cantabcapital dot com Keep up the good work! ** [shameless plug] Or if you are keen, enthusiastic, mostly competent, and looking for a well paid job where you get to do loads of Python SQLAlchemy -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sqlalchemy] matches_any: an extension to the Query object, and a HSTORE attribute access property
On Tue, Jan 7, 2014 at 11:14 AM, Philip Scott safetyfirstp...@gmail.comwrote: Hi folks, SQLAlchemy is at the heart of what we do at work (UK based hedge fund); so first of all a big thank you; I am not quite sure where we'd be without you. We would like to give back to the community as much as we can (I tried to get some of the developers on the company xmas present list this year but was too late.. cross your fingers for next year). We have extended SQLAlchemy in a few places, though it is quite intermingled with our domain specific stuff I keep an eye out for little snippets that might be useful to others. So here's a trivial one; take it or leave it (and feel free to think of a better name). Knowing my luck it already exists; though I have looked hard through the docs! class QueryEnhanced(Query): ''' Add a few extra bells and whistles to the standard Query object ''' def matches_any(self): ''' Returns true if your query would return 1 or more rows; false otherwise. The following two statements ask the same question; but matches_any is _much_ quicker on large tables: my_query.matches_any() my_query.count() != 0 ''' return self.session.scalar(select([self.exists()])) The other bit of technology we have that could be unpicked without _too_ much trouble is a sort of reverse CompositeProperty; many attributes of different types, including collections, out of one HSTORE column (with a sort of side-loaded instrumentation for mutation tracking that I think could have been done in a more idiosyncratic way). Paraphrasing a bit but you can do things like: class Animal(Base): data = Column(MutableDict.as_mutable(HSTORE)) colour = HsProperty(data, String) legs = HsProperty(data, Integer) discovered = HsProperty(data, Date) fun_facts = HsProperty(data, JSONEncoded(list)) 'colour', 'legs', 'discovered', and 'fun_facts' end up as keys in the HSTORE and the values are strings, integers, dates and lists on the python side but stored as strings in the HSTORE such a way that they can be CAST-ed in a server query [where possible]: session().query(Animal).filter(Animal.legs 2) and get a query like SELECT ... FROM animal WHERE CAST(animal.data - legs AS INTEGER) 2 You can also put an arbitrary JSONEncodable object in there too. Collections get converted to Mutable counterparts for change-tracking. In many ways it is similar to ColumnProperty except that - the properties are writable (and when written only trigger the relevant bits of the hstore to be updated). Also on object instances the values in HsProperties are fetched as part of the query; we lazily de-serialise them directly from the hstore dictionary. Before spend a couple of days removing our corporate clutter from that, getting permission to license it etc.. and posting either as a patch or extension I thought I would see if there is any interest (or if someone has already done it better?). It's implemented as a custom metaclass right now, but I think I might be able to do it fully with events. I would be very interested in this work. At my org we have a subset of the same idea that we're depending upon, but it's tied to an ancient SQLAlchemy version and we never took it all the way into the query space like that. That looks absolutely fabulous! -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sqlalchemy] matches_any: an extension to the Query object, and a HSTORE attribute access property
On Jan 7, 2014, at 2:14 PM, Philip Scott safetyfirstp...@gmail.com wrote: The other bit of technology we have that could be unpicked without _too_ much trouble is a sort of reverse CompositeProperty; many attributes of different types, including collections, out of one HSTORE column (with a sort of side-loaded instrumentation for mutation tracking that I think could have been done in a more idiosyncratic way). Paraphrasing a bit but you can do things like: class Animal(Base): data = Column(MutableDict.as_mutable(HSTORE)) colour = HsProperty(data, String) legs = HsProperty(data, Integer) discovered = HsProperty(data, Date) fun_facts = HsProperty(data, JSONEncoded(list)) 'colour', 'legs', 'discovered', and 'fun_facts' end up as keys in the HSTORE and the values are strings, integers, dates and lists on the python side but stored as strings in the HSTORE such a way that they can be CAST-ed in a server query [where possible]: session().query(Animal).filter(Animal.legs 2) and get a query like SELECT ... FROM animal WHERE CAST(animal.data - legs AS INTEGER) 2 You can also put an arbitrary JSONEncodable object in there too. Collections get converted to Mutable counterparts for change-tracking. In many ways it is similar to ColumnProperty except that - the properties are writable (and when written only trigger the relevant bits of the hstore to be updated). Also on object instances the values in HsProperties are fetched as part of the query; we lazily de-serialise them directly from the hstore dictionary. Before spend a couple of days removing our corporate clutter from that, getting permission to license it etc.. and posting either as a patch or extension I thought I would see if there is any interest (or if someone has already done it better?). It's implemented as a custom metaclass right now, but I think I might be able to do it fully with events. that’s a very nice pattern! It looks like you could do that strictly with Python descriptors, such as subclassing @hybrid_property, no ? Each property is just an expression against the “data” column, either python-side (lazily deserialize a key) or expression-side (do a postgresql expression for a certain key), and you can also trigger the mutable.is_changed() flag within the setter. I don’t think you’d need events. Code aside, if you can think of ways in which we as a company could support SQLAlchemy (bear in mind I am not in charge of the purse strings, but I can make a pitch on your behalf; we are still awaiting the fruits of our donation to the PyPy guys :).** Then do let me know. I don't check this email account all that regularly but my work address is my firstname.lastname at cantabcapital dot com Thanks ! I think what I’m usually looking for are people resources. Doc fixes and pull requests and such. If I could organize some significant batch of work as something that would work under a “grant” model, I’ll let you know. I haven’t figured out how to work that way, yet. signature.asc Description: Message signed with OpenPGP using GPGMail
Re: [sqlalchemy] SQLAlchemy 0.9.1 released
But I don't know why make this decision. Because where NULL will get nothing. And in 0.8.X version, I need to combine multiple condition according user input to one condition, so my code just like: cond = None for c in conditions: cond = c cond So in 0.9.X, the result will be something like: WHERE todo.id = 1 AND NULL, so I got nothing. Above code is totally wrong in 0.9.X. And I think maybe the old way makes sence. On Tue, Jan 7, 2014 at 10:42 PM, Michael Bayer mike...@zzzcomputing.comwrote: that’s this: http://docs.sqlalchemy.org/en/rel_0_9/changelog/migration_09.html#improved-rendering-of-boolean-constants-null-constants-conjunctions On Jan 7, 2014, at 4:13 AM, limodou limo...@gmail.com wrote: I found a problem in 0.9.1 version: in 0.8.x : print (Blog.c.id http://blog.c.id/==5) None blog.id = :id_1 But in 0.9.1: print (Blog.c.id http://blog.c.id/==5) None blog.id = :id_1 AND NULL So I don't know if it's a bug? On Tue, Jan 7, 2014 at 2:25 AM, Jonathan Vanasco jonat...@findmeon.comwrote: automap sounds neat! thanks! -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out. -- I like python! UliPad The Python Editor: http://code.google.com/p/ulipad/ UliWeb simple web framework: https://github.com/limodou/uliweb My Blog: http://my.oschina.net/limodou -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out. -- I like python! UliPad The Python Editor: http://code.google.com/p/ulipad/ UliWeb simple web framework: https://github.com/limodou/uliweb My Blog: http://my.oschina.net/limodou -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sqlalchemy] SQLAlchemy 0.9.1 released
On 08 Jan 2014, at 01:26, limodou limo...@gmail.com wrote: But I don't know why make this decision. Because where NULL will get nothing. And in 0.8.X version, I need to combine multiple condition according user input to one condition, so my code just like: cond = None for c in conditions: cond = c cond Why don’t you change the initial value to true() instead of None? If I read the documentation correctly that should work correctly in both SQLAlchemy versions. Wichert. -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sqlalchemy] SQLAlchemy 0.9.1 released
On Wed, Jan 8, 2014 at 3:31 PM, Wichert Akkerman wich...@wiggy.net wrote: On 08 Jan 2014, at 01:26, limodou limo...@gmail.com wrote: But I don't know why make this decision. Because where NULL will get nothing. And in 0.8.X version, I need to combine multiple condition according user input to one condition, so my code just like: cond = None for c in conditions: cond = c cond Why don’t you change the initial value to true() instead of None? If I read the documentation correctly that should work correctly in both SQLAlchemy versions. Even cond='' is correctly also, but I just think NULL is not a valid condition expression in SQL, so I think the old appoach maybe better. -- I like python! UliPad The Python Editor: http://code.google.com/p/ulipad/ UliWeb simple web framework: https://github.com/limodou/uliweb My Blog: http://my.oschina.net/limodou -- You received this message because you are subscribed to the Google Groups sqlalchemy group. To unsubscribe from this group and stop receiving emails from it, send an email to sqlalchemy+unsubscr...@googlegroups.com. To post to this group, send email to sqlalchemy@googlegroups.com. Visit this group at http://groups.google.com/group/sqlalchemy. For more options, visit https://groups.google.com/groups/opt_out.
