2026/03/21
Jump to navigation
Jump to search
|
Saturday, March 21, 2026 (#80)
|
|
References
|
- 10:21 The array-clades remain an unseemly mess. Just how deep shall I dive in trying to tidy them? I can at least (per decision a week or three ago) stop worrying about redundant interface-inheritance.
- 15:24 Did as little as I could do -- which ended up including a rename/move of
QValtoQVar\RO-- and the code is now back to its new-normal level of dysfunctionality. - 21:16 And now -- aha, yes, I do see the problem. The connection may be staying open at the logical level, but each SQL command is still being wrapped in an
echoclause, meaning each one is invoking a separate instance of the Engine, so... obviously the state is not going to be persistent. I think maybe I can fix this with just a flag, or at least it shouldn't require any new infra. - 21:40 Except... Q1: how is it still getting wrapped? The Command-object is being created with just the raw SQL...
$oCmd->AsString()produces "USE `htyp_mw`;- ...but
proc_get_status()shows the command asecho 'SHOW SCHEMAS;' | mariadb -hlocalhost -uroot -p<password redacted> - A1: It's happening because the next-link Server is
[WFe]Sys\Data\Engine\endpt\Server\MyMar\cMaria. Pretty sure I worked out a way to bypass the wrapping there...- Right: Upon return from
DoCommand($oCmd),$oCmd->QORunner()->GetIt()will be the endpoint Runner -- so I need to get that from somewhere and set it in all subsequent commands, and then modifyCommandsto use it if set (spoiler: already did that part). - Better, actually: just create the
Commandsqueue using thatRunner(instead of the next-linkServer), and then everything in the queue will be sent directly. - The only problem with all this is that now I find myself needing to pass the
Runneraround all over the place -- including the Engine Client, forSchemaExists(). I feel like there's got to be a better way... like, if that connection is now "open", shouldn't everything go there directly? Is there ever any reason not to send stuff to the end-Runner as long as the connection hasn't been closed? Or is this oversimplifying too much?
- Right: Upon return from