On Thu, Feb 19, 2009 at 1:16 PM, Edi Weitz edi@agharta.de wrote:
On Thu, Feb 19, 2009 at 12:04 PM, Vsevolod vseloved@gmail.com wrote:
- Last year I've sent a patch to CL+SSL, that allows it to work with
password-protected private key files. So the appropriate mention in the
docs
can be removed.
Thanks, didn't know that. But we'll have to change in ssl.lisp then, right? Do you have a patch for that?
I prepared the patch (it's a copy of how it works for me in the previous version). Basically, it's just an addition of a keyword parameter to the function call in ssl.lisp. It works without any problems with the previous version (for quite a long time now). With 1.0.0. I was as well able to create an acceptor and view the default page over SSL. NIL works for not encrypted key files. The diff it attached.
Additionally, I want to report some problems, which I've encountered while testing vesrion 1.0.0. 1. I get USOCKET:CONNECTION-REFUSED-ERROR on (hunchentoot-test:test-hunchentoot "http://localhost:4240") (my lisp is SBCL 1.0.20. I've updated bordeux_threads, chunga to the latest versions. flexi-streams v.1.0.7 I was already using).
2. If I do everything manually (setf *server* (make-instance 'hunchentoot:acceptor :port 4242)) (hunchentoot:start *server*) returns. If I browse to the site there's a default page there. But (hunchentoot:stop *server*) hangs.
Best regards, Vsevolod
On Thu, Feb 19, 2009 at 7:03 PM, Vsevolod vseloved@gmail.com wrote:
I prepared the patch (it's a copy of how it works for me in the previous version).
Thanks, it's in the dev version now. You forgot to update the docstrings and the documentation, though...
Additionally, I want to report some problems, which I've encountered while testing vesrion 1.0.0.
- I get USOCKET:CONNECTION-REFUSED-ERROR on
(hunchentoot-test:test-hunchentoot "http://localhost:4240") (my lisp is SBCL 1.0.20. I've updated bordeux_threads, chunga to the latest versions. flexi-streams v.1.0.7 I was already using).
- If I do everything manually (setf *server* (make-instance
'hunchentoot:acceptor :port 4242)) (hunchentoot:start *server*) returns. If I browse to the site there's a default page there. But (hunchentoot:stop *server*) hangs.
You're saying you get this even without SSL? Unfortunately, I don't have an installation of SBCL to test this right now. Hans, do you?
BTW, what's the value of bt:*supports-threads-p* for you?
On Fri, Feb 20, 2009 at 9:34 AM, Edi Weitz edi@agharta.de wrote:
On Thu, Feb 19, 2009 at 7:03 PM, Vsevolod vseloved@gmail.com wrote:
I prepared the patch (it's a copy of how it works for me in the previous version).
Thanks, it's in the dev version now. You forgot to update the docstrings and the documentation, though...
Sorry.
Additionally, I want to report some problems, which I've encountered
while
testing vesrion 1.0.0.
- I get USOCKET:CONNECTION-REFUSED-ERROR on
(hunchentoot-test:test-hunchentoot "http://localhost:4240") (my lisp is SBCL 1.0.20. I've updated bordeux_threads, chunga to the
latest
versions. flexi-streams v.1.0.7 I was already using).
- If I do everything manually (setf *server* (make-instance
'hunchentoot:acceptor :port 4242)) (hunchentoot:start *server*) returns. If I browse to the site there's a default page there. But (hunchentoot:stop *server*) hangs.
You're saying you get this even without SSL?
Yes. It's independent of SSL and occurred even before making my changes (on a vanilla version).
BTW, what's the value of bt:*supports-threads-p* for you?
T
On Fri, Feb 20, 2009 at 09:00, Vsevolod vseloved@gmail.com wrote:
On Fri, Feb 20, 2009 at 9:34 AM, Edi Weitz edi@agharta.de wrote:
On Thu, Feb 19, 2009 at 7:03 PM, Vsevolod vseloved@gmail.com wrote:
But (hunchentoot:stop *server*) hangs.
You're saying you get this even without SSL?
Yes. It's independent of SSL and occurred even before making my changes (on a vanilla version).
This is a usocket bug that I have a patch for. I will investigate this today and eventually commit a change for the problem. In the mean time, using an older usocket version helps (maybe from 6 months ago).
-Hans
On Thu, Feb 19, 2009 at 7:03 PM, Vsevolod vseloved@gmail.com wrote:
Additionally, I want to report some problems, which I've encountered while testing vesrion 1.0.0.
- I get USOCKET:CONNECTION-REFUSED-ERROR on
(hunchentoot-test:test-hunchentoot "http://localhost:4240") (my lisp is SBCL 1.0.20. I've updated bordeux_threads, chunga to the latest versions. flexi-streams v.1.0.7 I was already using).
Some more questions:
1. Did you also use the newest version of Drakma?
2. I guess this is a typo and you really meant "4242"?
3. Does it make a difference if you use "127.0.0.1" instead of "localhost"?
4. Does it make a difference if you run the server and the test in two different Lisp images?
Thanks, Edi.
On Fri, Feb 20, 2009 at 15:18, Edi Weitz edi@agharta.de wrote:
On Thu, Feb 19, 2009 at 7:03 PM, Vsevolod vseloved@gmail.com wrote:
Additionally, I want to report some problems, which I've encountered while testing vesrion 1.0.0.
- I get USOCKET:CONNECTION-REFUSED-ERROR on
(hunchentoot-test:test-hunchentoot "http://localhost:4240") (my lisp is SBCL 1.0.20. I've updated bordeux_threads, chunga to the latest versions. flexi-streams v.1.0.7 I was already using).
Just as a side note: I verified that hunchentoot-test:test-hunchentoot (sic) passes with no problems, in one image, on sbcl 1.0.21.18 (FreeBSD) multi-threaded. There currently is a problem with stopping and then starting the same acceptor. For some reason, this does not work, so for now, please always create a fresh acceptor object.
-Hans
On Fri, Feb 20, 2009 at 4:18 PM, Edi Weitz edi@agharta.de wrote:
On Thu, Feb 19, 2009 at 7:03 PM, Vsevolod vseloved@gmail.com wrote:
Some more questions:
- Does it make a difference if you run the server and the test in two
different Lisp images?
From this question I understand, that the server should be already running
by the time I make the test. Unfortunately, from the doc I've got the impression, that it's some kind of a clever test, that will start the server on it's own, will try to connect to it and will stop it afterwards (maybe it there's a point in doing that this way, because otherwise its not very useful -- it may be easier to browse to the web-page). If I start the server manually, it works. Sorry for the wrong input.
Thanks, Vsevolod
On Fri, Feb 20, 2009 at 9:58 PM, Vsevolod vseloved@gmail.com wrote:
maybe it there's a point in doing that this way
No, actually it's the other way around. The test suite should not be able to start and stop the server, it should just test it from the POV of a client. It shouldn't even be part of Hunchentoot but of Drakma and we will change that.
otherwise its not very useful -- it may be easier to browse to the web-page
The point is that it is faster and less error-prone to not do it manually.
If I start the server manually, it works.
That's good to hear.
Thanks, Edi.
-
Edi Weitz -
Hans Hübner -
Vsevolod