http://ayende.comAyende @ RahienCopyright (C) Ayende Rahien 2004 - 2021 (c) 202660Seems odd to have to run CheckStatementForUnboundedResultSet if data is available and set up an event handler for ValuesRefreshed that does the same kind of this later on. Why not model the process as adding a verification behavior to the statement scheduled to be performed when data becomes available (now or later). In fact, you might want to separate two different parts of the lifecycle: setup of verification behaviors associated with statements, and deep execution / inspection of those statements later on. http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment7http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment7Wed, 07 Jan 2009 09:50:08 GMTStephen, Parsing SQL and how NH does it are two separate issues. http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment6http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment6Fri, 02 Jan 2009 15:22:39 GMTI haven't used nhibernate to deeply, but I assume it has a provider model to it can understand different database implementations and how to at least generate query strings for it, if not also parse them.. I take it you aren't planning a providers model for the native query parsing (if the nhibernate providers aren't already expected to do this?). http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment5http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment5Fri, 02 Jan 2009 15:18:51 GMTParsing SQL is actually pretty easy, I already have to do this in order to get the pretty formatting working. http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment4http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment4Sat, 27 Dec 2008 12:01:09 GMTAyende, Parsing SQL looks like smell because it was generated by NHibernate and you could have access to it's object implementation... But I guess you are doing this because you can't actually access it via mechanism you have chosen (custom logger), right? http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment3http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment3Sat, 27 Dec 2008 08:30:12 GMTKonstantin, I am not aiming for 100% correctness here, but this can be dealt with using SQL parsing, which I am already doing. As for CountOfRows being nullable. This means the count of rows returned, so CUD statements has no such thing. http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment2http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment2Sat, 27 Dec 2008 05:06:29 GMTHey Ayende, I especially liked the trick with boolean addedAction in closure to make sure that you do checks until you add suggestion or realize it is not required. IndexOf you use to find limitKeword looks ugly. What will happen if RawSql contains column named "top_time_limit_offset"? Your code will not detect Unbounded Result Set in this case. Looks like a major bug in this feature to me. Why is your statement.CountOfRows nullable? Any query always deals with zero or more rows. http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment1http://ayende.com/3773/nh-prof-deep-dive-implemented-the-unbounded-result-set-warning#comment1Sat, 27 Dec 2008 04:06:12 GMT