MartinF, I say this as someone who has no stake in NH (I'm not part of the team, and I'm not using it. Only contributing re-linq, which NH LINQ builds upon):
The ORM Battle tests are biased. They come from a ORM maker. There's been enough time wasted arguing over that. So it's not a good idea to come here and say "you should do X because ORM Battle says Y". There may be some truth in them, but nobody can be bothered to look into their advertising suite and find out what matters. If you actually think that query performance is a priority, it should be because it is a limiting factor in your real-world app. If you're really interested, you can try to reproduce the tests that are bothering you in HQL. If they offer much better performance, maybe there is something wrong with NH LINQ caching, and it can't hurt to create an issue or even a patch! Only if caching cannot be fixed and performance is really a problem, we should consider the evilness that compiled queries are. Keep in mind that compiled queries are a solution for the problem that in LINQ to SQL, query building often takes longer than the DB roundtrip! A "where s.Id == id" test like the one you're referencing says nothing. Both caching AND compiling will gain much more for more complex queries. This test says nothing about real-world performance. OTOH, the tests measure the whole DB roundtrip, so there's little to learn about LINQ performance from that alone. On 28 Okt., 03:07, MartinF <[email protected]> wrote: > I dont want to start a battle. > Just wanted to know if it was (or will be) possible to compile (or > some other way) speed up the Linq queries so it becomes faster. > > I think NHibernate is perfect when it comes to mapping (using fluent > nhibernate) and everything else so far works fine, and the new lazy > properties feature is really nice. > Only thing i wish for is great linq support and fast queries that > generates good sql statements. > > NHibernate 3 A2 is not shown in the score cards, but if you look at > the comments here -http://ormbattle.net/index.php/nhibernate.html > there is a link (http://code.google.com/p/ormbattle/source/browse/ > Output.txt#375) to the new results which doesnt really show any > noticable difference (compared to old results - except for better > materialization). > > On Oct 27, 6:05 pm, Oskar Berggren <[email protected]> wrote: > > > > > But the published score card is dated july 2010 it seems. So maybe the > > code is updated but the "official" results are still based on NH 2.1? > > > /Oskar > > > 2010/10/27 Carlos cubas <[email protected]>: > > > > Yeah I'm sured, I was also skeptical but I browsed the source code, and > > > commit changes. Their web page needs to be updated. > > > > -Carlos > > > > Practice makes perfect, but if no one is perfect, why practice? > > > >> From: [email protected] > > >> Date: Wed, 27 Oct 2010 17:45:35 +0200 > > >> Subject: Re: [nhusers] Re: Compile Linq Query ? > > >> To: [email protected] > > > >> 2010/10/27 Carlos cubas <[email protected]>: > > > >> > @Fabio They are using Nhibernate 3 Alpha 2. > > > >> You sure? Some pages need to be updated then: > > > >>http://ormbattle.net/index.php/blog.html(lastentry on the subject > > >> from july 2010 says 3.0 alpha cannot be used) > > >>http://ormbattle.net/index.php/nhibernate.html > > > >> /Oskar > > > >> > @Martin These performance test are true but their interpretation is > > >> > highly > > >> > subjective. I see a greater value at looking at some of the queries > > >> > that > > >> > nhibernate linq provider is currently failing on according to their > > >> > site. > > >> > I'd rather see more of the linq impl test passing on this site than the > > >> > performance ones. > > > >> > -Carlos > > > >> > Practice makes perfect, but if no one is perfect, why practice? > > > >> > ________________________________ > > >> > Date: Wed, 27 Oct 2010 12:15:40 -0300 > > >> > Subject: Re: [nhusers] Re: Compile Linq Query ? > > >> > From: [email protected] > > >> > To: [email protected] > > > >> > have a look to the NH's version they are using. > > >> > btw we are not interested in such kind of battle. > > > >> > On Wed, Oct 27, 2010 at 12:03 PM, MartinF <[email protected]> wrote: > > > >> > Hello Stefan, > > > >> > thanks for your answer. > > > >> > I was just looking at some code using the Entity Framework which used > > >> > compiled queries and wondered if it would become possible to do > > >> > something similar with NHibernate. > > >> > I found this sitehttp://ormbattle.net/whichshows that most other > > >> > ORMs support Compiled Linq Queries and it seems to give a huge > > >> > performance boost to use them. > > > >> > Actually according to the tests on the site, Linq to Sql is only tiny > > >> > bit slower than NHibernate, and a lot faster when using Compiled Linq > > >> > Queries (almost 10x). > > > >> > The tests doesnt really simulate a real world scenario (executing the > > >> > same query over and over again from the same thread), but still it > > >> > indicates a performance difference in how it is implemented, and if > > >> > linq queries are cached in Nhibernate i dont understand why it doesnt > > >> > yield greater performance results in the tests compared to the other > > >> > ORMs. > > > >> > Martin :) > > > >> > On Oct 27, 9:33 am, Stefan Wenig <[email protected]> wrote: > > >> >> No plans AFAIK. Why would you want that? > > > >> >> LINQ to SQL has them because it is incredibly slow and does not cache. > > >> >> Compiled queries are a burden on the programmer: you need to manage > > >> >> the delegates (static members...), use this awful syntax, can only use > > >> >> up to 4 paramters (limit of Func<...>), and if you forget one > > >> >> parameter it will be silently captured and the query will just give > > >> >> you wrong results starting with the 2nd invocation (i.e., the unit > > >> >> test will work) > > > >> >> Did you actually measure the perfomance of NH LINQ and come to the > > >> >> conclusion that you need them? Because NH LINQ does cache, so > > >> >> transformation time for subsequent executions should be neglectable. > > > >> >> Cheers, > > >> >> Stefan > > > >> >> On 26 Okt., 02:28, MartinF <[email protected]> wrote: > > > >> >> > Hello, > > > >> >> > Is it possible to compile a linq query or will it become possible ? > > > >> >> > Martin- Hide quoted text - > > > >> >> - Show quoted text - > > > >> > -- > > >> > You received this message because you are subscribed to the Google > > >> > Groups > > >> > "nhusers" group. > > >> > To post to this group, send email to [email protected]. > > >> > To unsubscribe from this group, send email to > > >> > [email protected]. > > >> > For more options, visit this group at > > >> >http://groups.google.com/group/nhusers?hl=en. > > > >> > -- > > >> > Fabio Maulo > > > >> > -- > > >> > You received this message because you are subscribed to the Google > > >> > Groups > > >> > "nhusers" group. > > >> > To post to this group, send email to [email protected]. > > >> > To unsubscribe from this group, send email to > > >> > [email protected]. > > >> > For more options, visit this group at > > >> >http://groups.google.com/group/nhusers?hl=en. > > > >> > -- > > >> > You received this message because you are subscribed to the Google > > >> > Groups > > >> > "nhusers" group. > > >> > To post to this group, send email to [email protected]. > > >> > To unsubscribe from this group, send email to > > >> > [email protected]. > > >> > For more options, visit this group at > > >> >http://groups.google.com/group/nhusers?hl=en. > > > >> -- > > >> You received this message because you are subscribed to the Google Groups > > >> "nhusers" group. > > >> To post to this group, send email to [email protected]. > > >> To unsubscribe from this group, send email to > > >> [email protected]. > > >> For more options, visit this group at > > >>http://groups.google.com/group/nhusers?hl=en. > > > > -- > > > You received this message because you are subscribed to the Google Groups > > > "nhusers" group. > > > To post to this group, send email to [email protected]. > > > To unsubscribe from this group, send email to > > > [email protected]. > > > For more options, visit this group at > > >http://groups.google.com/group/nhusers?hl=en.-Hide quoted text - > > > - Show quoted text -- Zitierten Text ausblenden - > > - Zitierten Text anzeigen - -- You received this message because you are subscribed to the Google Groups "nhusers" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/nhusers?hl=en.
