Svante von Erichsen

The error is from printing, not from make-array. CL-USER> (defvar *w* (make-array 3 :fill-pointer t)) *W* CL-USER> *w* ERROR: Not a number!

I think the special value t for fill-pointer has to be handled at initialization. Also, 0-dimensional arrays are allowed, so check that there is exactly one dimension. ``` (defun make-array...

If I may interject — using `typecase` here is idiomatic. If I found a `cond` that checks a known type in each clause, I'd be inclined to change it to...

That is not a valid ordinary lambda list, so I guess it should be rejected.

I'd prefer ``` (if (plusp (length dbname)) ; works for NIL and "" the same dbname (getenv-default "PGDATABASE" user)) ```

I feel quite sure that the fix of command-line-arguments mentioned above should fix this issue. So, we need the new version of that. - You could for now compile yourself...

I'd actually expect the clojure-mode result (not alignment, but two levels of indentation to separate it from the body).

The command doesn't parse, but I'm not sure why. It seems that when you leave off the `&password=dbtestlocal` part, it would parse. According to the documentation in docs/command.rst, the password...

I'd say, it is maintained in bursts, but with long pauses.

Can't look deeper now, but trying to decode UTF-16 with UTF-8 would produce this problem, yes. Did you specify the --encoding option?