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

[hunchentoot-devel] https server - no shared cipher error
by Mac Chan 10 Jul '07

10 Jul '07

Hi, Have anyone successfully started a https server with the latest hunchentoot? For some reason I'm getting a "no shared cipher" error. I hoped that I'm not making stupid mistake with the ssl setup. # To generate a key $ openssl genrsa 1024 > server.key # To generate a Certificate Signing Request $openssl req -new -key server.key > server.csr # self-sign cert $openssl req -x509 -key server.key -in server.csr > server.crt (asdf:oos 'asdf:load-op :hunchentoot-test) (hunchentoot:start-server :port 4443 :ssl-certificate-file "server.crt") (setq hunchentoot:*catch-errors-p* nil) When I point firefox to https://localhost:4443/, a error dialog pops up: `Firefox and localhost cannot communicate securely because they have no common encryption algorithms.' I'm using lispworks 5.0.2 32-bit pro. This happens in both fedora core 6 and winxp. I used the precompiled ssl package found here : http://www.slproweb.com/download/Win32OpenSSL-0_9_8e.exe Here's the backtrace: SSL failure in #<COMM:SOCKET-STREAM 2009956F>: error:1408A0C1:SSL routines:SSL3_GET_CLIENT_HELLO:no shared cipher [Condition of type COMM:SSL-FAILURE] Restarts: 0: [ABORT] Quit process. Backtrace: 0: SIGNAL (#<COMM:SSL-FAILURE 2009C333> &REST NIL) 1: CONDITIONS::CONDITIONS-ERROR (:INVISIBLEP T COMM:SSL-FAILURE (:STREAM #1=#<COMM:SOCKET-STREAM 2009956F> :FORMAT-STRING "SSL failure in ~s: ~a" :FORMAT-ARGUMENTS (#1# "error:1408A0C1:SSL routines:SSL3_GET_CLIENT_HELLO:no shared cipher"))) 2: COMM::SSL-READ-OR-WRITE-BUFFER (#<COMM:SOCKET-STREAM 2009956F> T "ULT-DISPATCHERB@HC9 3: (METHOD STREAM:STREAM-FILL-BUFFER (STREAM:BUFFERED-STREAM)) (#<COMM:SOCKET-STREAM 2009956F>) 4: (METHOD STREAM:STREAM-READ-CHAR (STREAM:BUFFERED-STREAM)) (#<COMM:SOCKET-STREAM 2009956F>) 5: (METHOD STREAM:STREAM-READ-BYTE (COMM:SOCKET-STREAM)) (#<COMM:SOCKET-STREAM 2009956F>) 6: READ-BYTE (#<COMM:SOCKET-STREAM 2009956F> &OPTIONAL "<(EOF-ERROR-P T)>" :EOF) 7: READ-BYTE (#<CHUNGA:CHUNKED-IO-STREAM 2009BE8F> &OPTIONAL "<(EOF-ERROR-P T)>" NIL) 8: (METHOD FLEXI-STREAMS::READ-BYTE* (FLEXI-STREAMS::FLEXI-BINARY-INPUT-STREAM)) (#<FLEXI-STREAMS::FLEXI-BINARY-LATIN-1-IO-STREAM 2009BDF7>) 9: (METHOD STREAM:STREAM-READ-CHAR (FLEXI-STREAMS::FLEXI-LATIN-1-INPUT-STREAM)) (#<FLEXI-STREAMS::FLEXI-BINARY-LATIN-1-IO-STREAM 2009BDF7>) 10: READ-CHAR (&OPTIONAL "<(STREAM *STANDARD-INPUT*)>" "<(EOF-ERROR-P T)>" NIL :DONT-KNOW) 11: CHUNGA:READ-LINE* (#<FLEXI-STREAMS::FLEXI-BINARY-LATIN-1-IO-STREAM 2009BDF7> &OPTIONAL NIL) 12: HUNCHENTOOT::GET-REQUEST-DATA NIL 13: HUNCHENTOOT::PROCESS-CONNECTION (#<HUNCHENTOOT::SERVER 200A3EEF> 420) 14: (SUBFUNCTION MP::PROCESS-SG-FUNCTION MP::INITIALIZE-PROCESS-STACK) (:DONT-KNOW) 15: SYSTEM::%%FIRST-CALL-TO-STACK NIL Regards, -- Mac

2 2

[hunchentoot-devel] New app using Hunchentoot
by William Bland 27 Jun '07

27 Jun '07

Hi, Just wanted to say thanks for Hunchentoot and to let the list know I launched a new web app yesterday that uses it. In the app's first 24 hours, Hunchentoot has responded to more than 1.5 million http requests and has been absolutely rock-solid the whole time. The app, http://clutu.com is a collaborative crossword game. It's written 100% in CL, using sbcl on an ubuntu vps. In addition to Hunchentoot I used cl-ppcre, cl-who and mel-base. Thanks again for providing such an awesome web server to the Lisp community. Cheers, Bill. -- William Bland: http://abstractnonsense.com/ Lisp documentation: http://lispdoc.com/

4 3

[hunchentoot-devel] Custom default page
by Erick Lopez Carreon 23 Jun '07

23 Jun '07

Hello: I'm trying to point the Hunchentoot default page to my /index page, but i can't figure how. Here is my dispatch table: (setq *dispatch-table* (nconc (list 'dispatch-easy-handlers (create-static-file-dispatcher-and-handler "/favicon.ico" "/usr/share/vhosts/hdemo/favicon.ico") (create-static-file-dispatcher-and-handler "/main.css" "/usr/share/vhosts/hdemo/main.css")) (mapcar (lambda (args) (apply #'create-prefix-dispatcher args)) '(("/index" main-page) ("/admin" admin-page) ("/showpost" showpost) ("/editpost" editpost) ("/showtopic" show-topic-page))) (list #'default-dispatcher))) What i want is when i point my browser to: http://localhost:3000 I get the /index page not the default page (currently i get the Hunchentoot defaul page). Any clues would be appreciated. Thank you.

2 2

[hunchentoot-devel] url-encoding file names
by Mathias Dahl 19 Jun '07

19 Jun '07

Hi! I am encoding file names using the url-encode function in Hunchentoot. However, it encodes chars such as space and / as well, which makes at least my browser (Firefox) or web server not understand that they are indeed links to files I have shared. So I have this workaround: (defun replace-all (string part replacement &key (test #'char=)) "Returns a new string in which all the occurences of the part is replaced with replacement." (with-output-to-string (out) (loop with part-length = (length part) for old-pos = 0 then (+ pos part-length) for pos = (search part string :start2 old-pos :test test) do (write-string string out :start old-pos :end (or pos (length string))) when pos do (write-string replacement out) while pos))) (defun replace-some-encoded (file-name) ;; Some chars should not be encoded, apparently... (replace-all (replace-all file-name "+" "%20") "%2F" "/")) (defun url-encode-file-name (file-name) (replace-some-encoded (hunchentoot:url-encode file-name :utf-8))) The above gives URLs like these: http://myserver/Tommy%20K%C3%B6rberg%2FTommy%20K%C3%B6rberg%20-%20Stad%20i%… And without the workaround I get this: http://klibb.com/muuartist/Tommy+K%C3%B6rberg%2FTommy+K%C3%B6rberg+-+Anthem… I searched the doc for "file name" to see if there was some special method suited for encoding file names with path information but could not find one. Do I need the above workaround? Thanks! /Mathias

4 13

[hunchentoot-devel] PUT parameters missing?
by Jonathon McKitrick 17 Jun '07

17 Jun '07

I'm waaay behind in updating from 0.5.0, so yesterday I grabbed 0.11.1. I use a lot of PUT requests in my web app, and data is sent in parameters just like in POST. However, the latest version of hunchentoot does not appear to populate post-parameters in a PUT request. -- Jonathon McKitrick jmckitrick(a)reedlarkeygroup.com

3 9

[hunchentoot-devel] session cookie URI part gets in LSP expressions in Internet Explorer 7.0
by Andrei Stebakov 16 Jun '07

16 Jun '07

Since I couldn't find a mailing list for hunchentoot LSP port by Mac Chan I am posting here. The problem is that when I create a lsp page for hunchentoot and the page starts with <%(no-cache)%> <%(start-session)%> The page also has a bunch of inline expressions like <img src=<%=(get-image "HeaderTransp.gif")%>> When I go to Internet Explorer 7.0 (In Firefox it doesn't happen) and read the html source of the generated page I see following: <img src='?customize=16%3A5757A00F8834482F324324AED6F91D16'/lisphandler/images/HeaderTransp.gif> Basically all of my inline expressions like <%= (some-func) %> gets prepended by the '?customize=16%3A5757A00F8834482F324324AED6F91D16' When I hit the "refresh page" button it generates a good page without the artifacts. I am not sure if the problem on the LSP or hunchentoot side. Maybe no fix is required and I have to do something differently. Thank you, Andrew

3 3

[hunchentoot-devel] beginners output problem
by Andy Chambers 02 Jun '07

02 Jun '07

The send-reply function below is part of some code that transforms sexp's according to specified rules a-la-sxml from the scheme world. My problem is that when I evaluate the let form on its own, it returns "<HTML></HTML>" as expected. However, when I request the page served by the easy-handler, I get the following condition. I/O timeout reading #<SB-SYS:FD-STREAM for "a constant string" {B84B211}> I'm using sbcl on debian with hunchentoot just asdf-intalled as a local package. Can anyone suggest why the let form would evaluate ok on its own but not inside the easy-handler? Cheers, Andy (defun send-reply (&optional s &rest fragments) "Output the fragments to the specified stream" (labels ((descend (fragments &optional result) (cond ((null fragments) result) ((not (car fragments)) (descend (cdr fragments) result)) ((null (car fragments)) (descend (cdr fragments) result)) ((eq t (car fragments)) (descend (cdr fragments) t)) ((consp (car fragments)) (descend (cdr fragments) (descend (car fragments) result))) ((functionp (car fragments)) (funcall (car fragments)) (descend (cdr fragments) t)) (t (format s "~a" (car fragments)) (descend (cdr fragments) t))))) (descend fragments))) (define-easy-handler (test :uri "/test" :default-request-type :get) () (let ((out (with-output-to-string (s) (send-reply s '(#\< HTML (#\> "</" HTML #\>)))))) out))

2 1

[hunchentoot-devel] patches for Hunchentoot / Flexistream for Allegro MoDeRn lisp
by Gary King 02 Jun '07

02 Jun '07

Hi, I recently ran into a few minor glitches getting Hunchentoot to run under Allegro's Modern Common Lisp. Here are some patches (based on the darcs repos). The flexistream patch is the "important" one. The hunchentoot changes are just adding a declare ignore, fixing a typo in a docstring and rewiting some #+ #+ a b as #+ a #+ b (the former has a claim to elegance but the later is unambiguous and works regardless of the default settings in allegro). HTH, # Flexistreams diff -rN -u old-flexi-streams/packages.lisp new-flexi-streams/ packages.lisp --- old-flexi-streams/packages.lisp 2007-06-02 08:32:41.000000000 -0400 +++ new-flexi-streams/packages.lisp 2007-06-02 08:32:41.000000000 -0400 @@ -29,7 +29,7 @@ (in-package :cl-user) -(unless (find-symbol "STREAM-FILE-POSITION" :trivial-gray-streams) +(unless (find-symbol (symbol-name 'stream-file-position) :trivial- gray-streams) (error "You need a newer version of TRIVIAL-GRAY-STREAMS.")) (defpackage :flexi-streams # Hunchentoot diff -rN -u old-hunchentoot/server.lisp new-hunchentoot/server.lisp --- old-hunchentoot/server.lisp 2007-06-02 08:35:13.000000000 -0400 +++ new-hunchentoot/server.lisp 2007-06-02 08:35:13.000000000 -0400 @@ -159,6 +159,7 @@ but this works only on LispWorks - for other Lisps the key must not be associated with a password." ;; initialize the session secret if needed + (declare (ignorable port-provided-p)) (unless (boundp '*session-secret*) (reset-session-secret)) (let ((output-chunking-p t)) @@ -180,14 +181,20 @@ :dispatch-table dispatch-table :output-chunking-p (and output- chunking-p (not mod-lisp-p)) :input-chunking-p input-chunking-p - #-:hunchentoot-no-ssl #-:hunchentoot-no-ssl - :ssl-certificate-file (and ssl- certificate-file - (namestring ssl-certificate-file)) - #-:hunchentoot-no-ssl #-:hunchentoot-no-ssl - :ssl-privatekey-file (and ssl- privatekey-file - (namestring ssl-privatekey-file)) - #-:hunchentoot-no-ssl #-:hunchentoot-no-ssl - :ssl-privatekey-password ssl- privatekey-password + #-:hunchentoot-no-ssl + :ssl-certificate-file + #-:hunchentoot-no-ssl + (and ssl-certificate-file + (namestring ssl-certificate-file)) + #-:hunchentoot-no-ssl + :ssl-privatekey-file + #-:hunchentoot-no-ssl + (and ssl-privatekey-file + (namestring ssl-privatekey-file)) + #-:hunchentoot-no-ssl + :ssl-privatekey-password + #-:hunchentoot-no-ssl + ssl-privatekey-password :mod-lisp-p mod-lisp-p :use-apache-log-p (and mod-lisp-p use-apache-log-p) :read-timeout read-timeout diff -rN -u old-hunchentoot/test/test.lisp new-hunchentoot/test/ test.lisp --- old-hunchentoot/test/test.lisp 2007-06-02 08:35:13.000000000 -0400 +++ new-hunchentoot/test/test.lisp 2007-06-02 08:35:13.000000000 -0400 @@ -150,7 +150,7 @@ (log-message :info "Oops info") (log-message :debug "Oops debug") (error "An error was triggered on purpose. Check your ~ -Apache error log. Up to 12 messages where logged depending on ~ +Apache error log. Up to 12 messages were logged depending on ~ the Apache log level set in httpd.conf.") (:html (:body "You'll never see this sentence...")))) @@ -580,3 +580,4 @@ ("/hunchentoot/test/files/" send-file) ("/hunchentoot/test" menu))) (list #'default-dispatcher))) -- Gary Warren King, metabang.com Cell: (413) 885 9127 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM

2 1

[hunchentoot-devel] Custom types for easy handler.
by Timothy Schaeffer 01 Jun '07

01 Jun '07

Hello all, I propose that the types for easy-handlers be extensible by the users. I have pieces of markup which are used throughout our site which collect information from several fields into a class (in our case tied to a database with clsql). The names of the fields are related, but diffierent from the lists, arrays and hashes of the easy-handlers, Dates and addresses are two good examples, but there are others. I would like to be able to teach the easy handler to collect these for me and pass them as parameters. This cannot be done with functions as the :parameter-type because the functions passed as :parameter-type do not take the parameter name. There are two ways I think this could be done: - The easier way: add a :paramater-reader arg to the parameter description, and if passed use it in lieu of the ((post|get)-)?parameter functions in compute-parameter. The following might do it (untested): (defun compute-parameter (parameter-name parameter-type request-type &optional parameter-reader) ; <===== "Computes and returns the parameter\(s) called PARAMETER-NAME and converts it/them according to the value of PARAMETER-TYPE. REQUEST-TYPE is one of :GET, :POST, or :BOTH." (when (member parameter-type '(list array hash-table)) (setq parameter-type (list parameter-type 'string))) (let ((parameter-reader (or parameter-reader ; <==== (ecase request-type (:get #'get-parameter) (:post #'post-parameter) (:both #'parameter))) ) (parameters (and (listp parameter-type) (case request-type (:get (get-parameters)) (:post (post-parameters)) (:both (append (get-parameters) (post-parameters))))))) (cond ((atom parameter-type) (compute-simple-parameter parameter-name parameter-type parameter-reader)) ((and (null (cddr parameter-type)) (eq (first parameter-type) 'list)) (compute-list-parameter parameter-name (second parameter-type) parameters)) ((and (null (cddr parameter-type)) (eq (first parameter-type) 'array)) (compute-array-parameter parameter-name (second parameter-type) parameters)) ((and (null (cddddr parameter-type)) (eq (first parameter-type) 'hash-table)) (compute-hash-table-parameter parameter-name (second parameter-type) parameters (or (third parameter-type) 'string) (or (fourth parameter-type) 'equal))) (t (error "Don't know what to do with parameter type ~S." parameter-type))))) (defun make-defun-parameter (description default-parameter-type default-request-type) "Creates a keyword parameter to be used by DEFINE-EASY-HANDLER. DESCRIPTION is one of the elements of DEFINE-EASY-HANDLER's LAMBDA-LIST and DEFAULT-PARAMETER-TYPE and DEFAULT-REQUEST-TYPE are the global default values." (when (atom description) (setq description (list description))) (destructuring-bind (parameter-name &key (real-name (compute-real-name parameter-name)) parameter-type init-form request-type parameter-reader ) ; <==== description `(,parameter-name (or (and (boundp '*request*) (compute-parameter ,real-name ,(or parameter-type default-parameter-type) ,(or request-type default-request-type))) ,init-form)))) (This just occurred to me: request-type could be a function, with an obvious change to the ecase. This is a little more confusing but consistent with what parameter-type does---so it has the force of precedent---and requires a smaller change to the code.). - The extensive way: Change the cond in compute-parameter to a switch via a hash table, whose values are the conversion handlers (or a cons with reader and converter). It may need two hash tables: one for simple type names (type is a symbol) and one for aggregates, (type is a cons, and switch on the first element). Create a macro, define-easy-type, or something, to populate the tables, and use it within easy-handlers.lisp to populate it with those which are now hardcoded in the cond and case switches. Users can then use the macro to define their own types. - Another way: like above, but use methods instead of hash tables. This seems possible, but I'd have to think some more about it. BTW: why the checks for null CDDRs and CDDDDRs in compute-parameter? Is this just nice, or are they essential? If your interested in the second, I will work something up and send in a patch. I do not think it would break any existing code which did not peak too far into the easy-handlers stuff. Let me know what you think. Tim S P.S.: Hunchentoot is a pleasure to work with!

3 2

[hunchentoot-devel] Re: Hunchentoot REST framework
by Edi Weitz 29 May '07

29 May '07

Hi, On Tue, 29 May 2007 15:35:58 +0200, Nicolas Lamirault <lam(a)tuxfamily.org> wrote: > i'm re-working on this framework and i've got a few problems ... I'd prefer if we could continue this discussion on the mailing list - see Cc. > i try to get parameters when a make a PUT request : > > $> curl -X PUT -d name=nicolas http://localhost:8888/ht-rest/test/user/lam@foo.org If you use curl like this, it will send the parameters as application/x-www-form-urlencoded in the request body. I can't see how it would make sense to use a PUT request for this. In a PUT request, the request body is supposed to be the entity that is to be stored under the request URI. > on logs i have this : > > [2007-05-29 15:30:19 [INFO]] Put User > [2007-05-29 15:30:19 [INFO]] PUT parameters NIL Where do these log entries come from? > and i create a function : > > (defun rest-put-parameter (name) > "Returns the value of the PUT parameter named by the string NAME as > a string or NIL if there ain't no PUT parameter with this name." > (hunchentoot:aux-request-value (make-symbol name))) > > but my function doesn't work ... > I think my mistake comes from i don't really understand how use the > aux-request-value function. Yes, it looks so. AUX-REQUEST-VALUE is meant so you can /store/ arbitrary values associated with the request and later retrieve them. I have no idea what REST-PUT-PARAMETER is supposed to do or what a "PUT parameter" is. > have u got any idea for this problem ? I don't understand the problem... Did you look at CL-WEBDAV to see how it handles PUT requests as I told you? Cheers, Edi.

1 0