In that case we behave the same as when the variant is known to be :pgdg.
The only other supported variant at this time being :redshift, that should
be okay.
The way we manage and then fetch the SQL queries embedded in the pgloader
binary we should really take the unix-namestring rather than the
native-namestring. Of course, this only matters when the host OS is NOT
unix, which is why this bug existed for so long.
* Improve documentation with command lines and defaults.
The advanced documentation coverage skipped the simple approach and didn't
explain fully what to do with the sample commands.
Fixes#1413.
* Fix docs indentation in the MySQL connection string options.
* Desultory docs and docs config fixes.
* SBCL compiler notes should not be fatal to pgloader.
The compile function returns warnings-p and failure-p values, use that to
decide if the code could be compiled, and only signal a condition when it
has been fatal to compiling the code at run-time.
The SBCL compiler is getting smarter at removing unreachable code, and it
looks like pgloader is producing some unreachable code from parsing the user
provided commands.
* Let's make the code look like actual lisp code now.
* Another fix.
* Improve condition handling and simplify processing of compile values.
We don't need to react to any condition signaled from inside pgloader, only
to errors and serious-conditions (an error is a serious-condition). With
that, we can just ignore the compiler warnings and style notes.
* Fix the handler-bind to only consider serious-conditions too.
* Capture compiler output as log it as a debug level message.
* Fix previous attempt.
* Improve capturing of the compiler output (include summary).
* Actually call the new compile function in all places.
Co-authored-by: Dimitri Fontaine <dimitri@citusdata.com>
cl+ssl::libcrypto is also read at startup. If not properly closed in the
hooks, together with libssl, libcrypto will be loaded at startup and if
the first cl+ssl file alternative fails, the debugger is invoked.
* Fixes#1370
