- tbnl-devel@common-lisp.net - mailman3.common-lisp.net

[hunchentoot-devel] [PATCH] handle ACCEPTOR-SHUTDOWN-P correctly
by Desmond O. Chang 21 Mar '11

21 Mar '11

After stopping an acceptor with STOP, one can't start it again with START, because the START method doesn't reset the value of acceptor-shutdown-p to NIL, so the listener returns/quits instantly. Also, the default value of ACCEPTOR-SHUTDOWN-P should be T, because every acceptor should be closed when it is created. See [1] & [2] for details. [1] http://article.gmane.org/gmane.lisp.lib.tbnl.general/2469 [2] http://article.gmane.org/gmane.lisp.lib.tbnl.general/2868 --- acceptor.lisp | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/acceptor.lisp b/acceptor.lisp index ba90e2e..168c9b9 100644 --- a/acceptor.lisp +++ b/acceptor.lisp @@ -120,7 +120,7 @@ connections.") :documentation "Number of pending connections allowed in the listen socket before the kernel rejects further incoming connections.") - (acceptor-shutdown-p :initform nil + (acceptor-shutdown-p :initform t :accessor acceptor-shutdown-p :documentation "A flag that makes the acceptor shutdown itself when set to something other than NIL.") @@ -266,6 +266,7 @@ they're using secure connections - see the SSL-ACCEPTOR class.")) ;; general implementation (defmethod start ((acceptor acceptor)) + (setf (acceptor-shutdown-p acceptor) nil) (start-listening acceptor) (let ((taskmaster (acceptor-taskmaster acceptor))) (setf (taskmaster-acceptor taskmaster) acceptor) -- 1.7.4.1

3 3

[hunchentoot-devel] hunchentoot-mp
by Haris 20 Mar '11

20 Mar '11

Hi. Where do I download hunchentoot-mp package ? Thanks

4 4

[hunchentoot-devel] Installing Hunchentoot on SBCL, Ubuntu 10.X
by Oleg Sivokon 19 Mar '11

19 Mar '11

Hello. Please forgive me asking questions with probably obvious answers. I'm new to the Lisp world in general, and to using Hunchentoot too. Since I'm new, I had a lot of problems installing Hunchentoot... I had successfully did it on two Ubuntu 10.4 machines, one 32 bit and another 64 bit both running last SBCL available from Ubuntu PPA. I had also tried to record my experience so I could explain it to anyone who asked. I'm effectively writing a mock-up server to do testing for the project that otherwise runs in a different environment, yet, at some point I would like other people to be also able to replicate my steps on installing the server and running my code. Major problems I encountered: - ASDF does not update... (I know, you have nothing to do with this, yet whenever the manual on the site mentions ASDF, it would be great if it would take in account that ASDF not only may do things wrong, it most probably will...). Long story short, I've tried to update ASDF following manual on their site - never worked due to some missing methods or just reverting to the last version w/o any notion. I've also tried building SBCL from sources. The last version in development comes with latest ASDF version (2.0), yet the installation instructions are bizarre and contradict themselves... At the most successful point I could install what compiles from trunk, but in order to run that version as root, I would need to specify the sbcl.core on launch. I could also install Hunchentoot into this version of SBCL, but it had problems with UFFI (separate story below). Various other packages in this mode would report missing files or non-existing functions and so on. It proved to be unusable. - ASDF-INSTALL, similarly, trying to install from archive or from site (clicky) were subsequently unsuccessful. Having only clean SBCL install, ASDF-INSTALL would break at random trying to solve dependencies. I couldn't find the pattern. By trial and error I found, that the combination of having CL-PPCRE, USOCKET, CLOSER-MOP and CXML (!) would most of the time lead to that Hunchentoot would be able to find all dependencies during install. This happened 2 times out of maybe 8... - There's some package, which during installation would break the display in console... (the console will start displaying random non-printable characters) It would also stop during compile offering different restarts, where you are forced to play pitch-n-toss :) And if you are lucky, then the console will recover. - Finally, the most sure way I found was to just (require :hunchentoot) using ASDF v 1 and resolve missing dependencies by hand. It would sometimes break when trying to install CL-PPCRE, presumably due to another dependency having something to do with Unicode, but it succeeded more often then failed. - UFFI - for all I tried it failed to compile and install properly always. It looks like the errors are in *insignificant* tests, yet for an unexperienced user this looks scary (mistyped pointers?!) :) Often the errors when compiling this package will also break the installation started by ASDF - it will not offer to skip the failed compilation, or so I understood what happened. (I also saw some messages regarding omission of "dead" code, while compiling this package - this also looks suspicious, why would anyone release such code, and may there be any connection between that, and that the code failed to compile in the end?) Some weird things I saw during installation: an attempt was made to install SLIME. It somehow got into dependency list of Hunchentoot... I've encountered several times notion of 64-bit processor architecture in different libraries related to Common Lisp as "PPC" - I never heard of this abbreviation before in this context, (no pun intended), what does it mean? This was one of really confusing places when reading the documentation (I would assume it to be Apple Motoralla processor from 10 years ago). For whatever other arcane reason Ubuntu / Debian tech papers call supported 64-bit processors AMD86 or AMD8664 (some of them having nothing to do with AMD!). Now, my question is, is my experience something you would expect a user to undergo? Is SBCL an experimental brunch of Common Lisp, and, would I want to develop in more stable environment, do I have to switch to another version of CL? (Back in the days I've used CMUCL on Windows, albeit very shortly, but I don't recall that many problems...) Or are Hunchentoot and SBCL not really compatible, and the attempt to put them together makes it so difficult? Of course, most chances are I am doing something wrong :) Yet it would be nice if someone more experience commented. Oh, and btw, I've managed to then install CL-GD and use it with Hunchentoot, for that I had to, again, delete some shared libraries in system folders and rewrite Make file... It worked, as I said, but, again, I've a feeling I do things the wrong way and it shouldn't be this difficult... And, one last thing, if there are bugs filed against the server, is it possible to view them somewhere? I've found few bug-trackers during my setup exercises :) however, I'm not sure which is an official one. Thank you.

7 10

[hunchentoot-devel] turning off logging, and socket backlogs (patches included)
by JTK 15 Mar '11

15 Mar '11

[This is all in a recent bknr svn hunchentoot running on opencml and sbcl on a mac 10.6] 1. Logging: I've been playing with running hunchentoot under heavy loads using easy-acceptor, and running against apache-bench (ab) I noticed that logging to *error-output* often causes socket failures under heavy load, at least in SLIME. Is there a reason why with-open-file-or-console in log.lisp couldn't use a NIL pathname to simply not log, and pathname=T would log to *error-output*, which is what NIL presently means? Now, there seems to be no way to turn off logging (besides subclassing acceptor). 2. Backlog in listener: I found that in the start-listening method in acceptor.lisp, there is no BACKLOG set for usocket:socket-listen. I think that others have pointed this out as well. The default apache backlog seems to be 100 to 512 [1,2,3] though the default on a macintosh seems silently limited to 128 [4]. In the usocket routines used by hunchentoot backlog seems to be only 5. This seems two orders of magnitude smaller than what is standard for web servers. I tried putting in :BACKLOG 50 into the start-listening method in acceptor.lisp and it eliminated most failures of ab. Is there any reason why acceptor objects shouldn't have a new backlog slot default value in them much larger than 5, which should be used in the start-listening method? The appended patches implement both changes, if these suggestions are judged useful. The backlog is 50. I hope they're done correctly; I'm not experienced in submitting patches. John [1] http://groups.google.com/group/modwsgi/browse_thread/thread/b6d66d3fe5a53d2c [2] http://weblog.pigfoot.org/pigfoot/2007/07/05/whats-the-maximum-value-of-bac… [3] /usr/include/apache2/mpm_common.h:#define DEFAULT_LISTENBACKLOG 511 [4] http://developer.apple.com/library/mac/#documentation/darwin/reference/manp…

2 1

[hunchentoot-devel] sessions
by Haris 14 Mar '11

14 Mar '11

Hi. Could someone give me a simple example of using sessions ? I have a simple web application and want to add user registration so that the server knows who submitted some form. Thanks

1 0

[hunchentoot-devel] encoding
by Haris 09 Mar '11

09 Mar '11

Hi. How to set Hunchentoot encoding to utf-8 ? Thanks

2 1

[hunchentoot-devel] encoding
by Lou Vanek 09 Mar '11

09 Mar '11

Haris, see, https://sites.google.com/site/sabraonthehill/home/exploring-quicklisp-packa… for an answer.

1 0

[hunchentoot-devel] Hunchentoot on CLISP
by Andrew Pennebaker 03 Mar '11

03 Mar '11

The "New York version" (hunchentoot:start (make-instance 'hunchentoot:acceptor :port 4242)) creates a stalling web server, never loading, never error'ing. When I remove the code :ready-only t from line 377: (when (usocket:wait-for-input listener :timeout +new-connection-wait-time+) The New York version works fine. Specs: - Hunchentoot 1.1.1 - Quicklisp - CLISP - Mac OS X

4 10

[hunchentoot-devel] Robustness change committed
by Hans Hübner 01 Mar '11

01 Mar '11

Hi, I committed a (modified) patch submitted by Dan Weinreb that makes Hunchentoot more resilient to high load conditions by limiting the number of request handler threads that can be running at any time. If you are running a site with a lot of traffic, you'll want to be aware of this change. Documentation for this feature is not yet available. -Hans ---------- Forwarded message ---------- From: BKNR Commits <bknr(a)bknr.net> Date: Thu, Feb 10, 2011 at 6:22 PM Subject: [bknr-cvs] hans changed trunk/thirdparty/hunchentoot/ To: bknr-cvs(a)common-lisp.net Revision: 4645 Author: hans URL: http://bknr.net/trac/changeset/4645 Patch from Dan Weinreb: limit the maximum number of request handler threads that a taskmaster may create. Support a new SOFT argument to the STOP function that makes Hunchentoot wait until all currently outstanding requests have been processed before returning. Add a favicon.ico to the standard file tree. Refactorings in the START-OUTPUT function to separate out the actual header sending from the request processing part. The new SEND-RESPONSE function does the I/O part of sending the response header (and possibly content) and also does the logging. Used by the thread count limiting code to send a 503 message if no more threads can be created. U trunk/thirdparty/hunchentoot/acceptor.lisp U trunk/thirdparty/hunchentoot/compat.lisp U trunk/thirdparty/hunchentoot/headers.lisp U trunk/thirdparty/hunchentoot/taskmaster.lisp U trunk/thirdparty/hunchentoot/util.lisp A trunk/thirdparty/hunchentoot/www/favicon.ico Change set too large, please see URL above _______________________________________________ bknr-cvs mailing list bknr-cvs(a)common-lisp.net http://common-lisp.net/cgi-bin/mailman/listinfo/bknr-cvs

2 1

[hunchentoot-devel] Hunchentoot examples aren't working for me
by JTK 28 Feb '11

28 Feb '11

Hello, I'm on a 10.6 mac, running sbcl and ccl/openmcl inside slime. I tried installing the latest svn hunchetoot from bknr-svn, replacing my old version of h'toot and the supporting libraries with the whole tree. Then I followed the instructions for the example server: (hunchentoot:start (make-instance 'hunchentoot:acceptor :port 4242)) (hunchentoot:define-easy-handler (say-yo :uri "/yo") (name) (setf (hunchentoot:content-type*) "text/plain") (format nil "Hey~@[ ~A~]!" name)) ==> SAY-YO Now the main page http://127.0.0.1:4242/ is OK - I see 'Welcome' and the doc tree. But http://127.0.0.1:4242/yo gives "Resource /yo not found" Similarly, (asdf:oos 'asdf:load-op :hunchentoot-test) loads But http://127.0.0.1:4242/hunchentoot/test give 'resource not found.' Is there anything obvious that I'm doing wrong? (Also, for some reason sbcl didn't like the fact that babel didn't have the positive-fixnum type defined, so I defined it. CCL was OK with babel as it was.) Thanks, John Klein

3 4