On Fri, May 07, 2021 at 09:06:31AM -0000, Stuart Henderson wrote:
> On 2021-05-05, T. Ribbrock <emga...@gmx.net> wrote:
[...]
> This is not a bad place to ask. Your description is good but anyone
> looking into what's up will want to test, so if you could include
> the test tools and a description of setup needed to reproduce that
> would help. Including the tcpdump traces would help too. Don't worry
> about the mail being long.
thank you! In that case, I'll spend some more time in providing the
items you mentioned to get them even more generic - currently, the
test config is based on my main server and has a few, let's say
"historic idiosyncracies" in it. I have always installed Nextcloud
manually up until now and I've already noticed some differences
between the Nextcloud package-readme and my own config. Not that I think
that this will have any significant influence on the issue at hand, but
I want this to be as easy to reproduce as possible.
I should be able to post them over the next few days (bit strapped for
time).
[...]
> > The Qt-client sends the http-headers first in one TCP-segment (I'm not
> > too good on terminology...). Once that has happened, httpd already sends
> > back the 401 - and *then* the Qt-client sends the XML-body in a second
> > TCP-segment, causing the "400 Bad Request" response (I presume because
> > httpd is expecting new headers at this point, not a content body).
> It makes no difference to the HTTP protocol whether headers and body are
> in separate TCP segments, but some software may handle things wrongly.
> httpd uses libevent and it wouldn't be the first time libevent-based
> software has problems with data in separate TCP segments (I have a
> feeling we might have had a problem with ftp-proxy related to this
> but can't find any details, perhaps it was never fixed),
That's what I thought as well, but I'm simply not deep enough in the
matter to make that call. My skills are just about good enough to read
through the httpd sources and with the help of the debug output to get
*some* idea of the flow, but that whole libevent-part is beyond me.
[...]
> Pretty sure it will be on the httpd side.
Ok, then I can at least already tell the SailfishOS community member who
helped me that he can put this on hold for now on his side. That's something!
Regards,
Thomas
--
-----------------------------------------------------------------------------
Thomas Ribbrock http://www.ribbrock.org/
"You have to live on the edge of reality - to make your dreams come true!"
