José Luis Domingo

Had this working very easily for a no-name ESP32 development board sourced from the lowest bidder through Aliexpress. Will post here the whole "git diff" against the current project version,...

Attaching the log file from a pristine "lnav" session and no $HOME/.lnav/ directory for clarity: [debug-sql.log](https://github.com/user-attachments/files/23657190/debug-sql.log)

Yes, that's right, exactly one result: ``` ✔ SQL Result: type=table; name=syslog_copy; tbl_name=syslog_copy; rootpage=2; sql=CREATE TABLE syslog_copy( log_line INT, log_time NUM, log_level TEXT, log_hostname TEXT, log_msgid TEXT, log_pid TEXT, log_pri...

OK, so took the time to investigate this problem using opencode.ai (using the free Grok Code Fast 1 model), to give the "agentic technology a try". Note my C++ knowledge...

Adding the diff that the tool created. Double checked that: * A pristine "lnav" tree cloned from github, and compiled, fails to populate the .db file when doing ";.save" (neither...

Confirmed it is working with your changes. Regarding the tables created from ":create-logline-table" and ":create-search-table", they are neither kept in the session (:export-session-to) nor saved when doing a ";.save", are...

Confirmed now ":create-search-table" commands are exported in the session file. And while I fail to understand the problems with the "logline stuff", it is handy (and much easier than :create-search-table)...

Thanks, but it doesn't quite get me something I am too comfortable with. The "ordered-by-time": false option in the format specification does as described, however the performance impact seems to...

Other than the supposed inefficiency described in my earlier comment, I am good with the handling of "out of order" messages, although what I did in the end is through...

Both the "sqlite3" binary as the libraries used to compile / link the "lnav" binary are the same version (running on a Fedora OS 42): ```shell $ sqlite3 --version 3.47.2...