FLOOD_WAIT errors are not propagated

[roboslone]

I believe the issue is missing break here.

Also, what would happen in case of 5xx error?

[BenWiederhake]

A few lines down:

if (strncmp (error, "FLOOD_WAIT_", 11)) {
          // …
          wait = 10;

It seems it is handled, but apparently with an insufficient wait time or something like that.

[roboslone]

The thing is, tg never sends those to my wrapper (through a unix socket). But if I add that missing break and recompile I get messages with FLOOD_WAIT_ prefix in error field. That gives me total control over flood.