Re: [Firebird-devel] Reading from batch
On 1/3/22 16:29, Jiří Činčura wrote: Hi *, Is there a way to read from a batch? I.e. executing `insert into foo (x) values (?) returning x` with bunch of parameters in a batch a then getting the values of `x`. Planned for next version, that will be another interface. Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
Re: [Firebird-devel] Different commands in a batch
On 1/3/22 16:23, Jiří Činčura wrote: Hi *, Executing different commands in a batch is not supported, right? Aka, it's only single command with multiple sets of parameters. Yes. Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
[Firebird-devel] Reading from batch
Hi *, Is there a way to read from a batch? I.e. executing `insert into foo (x) values (?) returning x` with bunch of parameters in a batch a then getting the values of `x`. -- Mgr. Jiří Činčura https://www.tabsoverspaces.com/ Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
[Firebird-devel] Different commands in a batch
Hi *, Executing different commands in a batch is not supported, right? Aka, it's only single command with multiple sets of parameters. -- Mgr. Jiří Činčura https://www.tabsoverspaces.com/ Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
[Firebird-devel] New development re. "contracted" collations and new QA
Hi all, I'd like inform you about new developments regarding: a) performance issues with partial key lookup with "contracted" collations. We've got sponsor to implement Adriano's solution using 3.0.8 code as primary target. The development and testing should happen in next three months. If everything will go well, the pull request with verified solution would be ready for review in early April (or end of March). We hope that it would be accepted for merge into 3+ branches. b) New QA based on pytest We have finished conversion of the whole tests suite from fbtest to new system (incl. changes in fbt-respository up to 31th December). About two hundred of tests still wait for reimplementation or decission about their fate (mostly those that depend on changes to firebird configuration or external tool like encryption plugin etc.), but the majority of test suite is functional and robust. This means that phase one is finished and we entered phase two - the real transition to new QA system. For some time, both QA systems could / would be operated in parallel, with a target to fully deprecate the old system for testing FB 3, 4 and 5 this summer (or sooner if possible). At this point, development or update of tests in old system is possible but discouraged to streamline and speed up the transition. best regards Pavel Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
Re: [Firebird-devel] p_batch_cs on the wire
On 1/3/22 12:07, Jiří Činčura wrote: ULONG p_batch_errors; // error's recnums Numbers of records with error. Used when too many errors took place. For each error server sends number of message for which execution of SQL statement failed. And yes - such number is 4-byte integer, data is sent after an array of detailed error info. What "too many errors" mean in real world? Desired to user number of errors with full details is set by TAG_DETAILED_ERRORS in batch parameters. But (like almost all our parameters) server does not guarantee exactly following requested value. Later implementation details follow - they may be changed at any time. If requested value > 256 actual value is 256. Default is 64. Exact limit of detailed errors does not affect CS parsing - all needed info is present internally. Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
Re: [Firebird-devel] p_batch_cs on the wire
> ULONG p_batch_errors; // error's recnums > > Numbers of records with error. Used when too many errors took place. For > each error server sends number of message for which execution of SQL > statement failed. And yes - such number is 4-byte integer, data is sent > after an array of detailed error info. What "too many errors" mean in real world? -- Mgr. Jiří Činčura https://www.tabsoverspaces.com/ Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel