added tests that exhibit issue #980 + fix; also fix #979 #993
Conversation
…rams"; supply "body".
…lization is NOT optional.
…ms() make sure newly added parameters get _decode()d upon repeated internal calls to _build_params().
|
... this has sat for far, far, far too long - I'm almost embarassed to comment on it it's been that long - but it appears to contain some important bugfixes. I'm sorry this hasn't been acted upon for some time. I see that a fix for the content length being miscalculated with UTF-8 got merged, at least, but there's other worthwhile stuff in here. For instance, returning an error if deserialisation fails is obviously sensible - need to check if current devel really does carry on with request processing if deserialisation fails, I find that hard to believe. So, yes - sorry that this PR has sat unattended to for so long. I'm trying to catch up on outstanding PRs and bug reports as fast as I can now I've taken over as D1 release manager. I think I'll get a new developer release out to CPAN soon with the stuff I've merged recently, then re-visit this PR to review it in detail and get it merged. |
|
biennial ping :-/ |
content length is miscalculated when the app has no 'charset' setting.
adding the pragma 'use bytes' fixes the issue.
#980
fix #979 too.