From jasondunsmore at gmail.com Thu Aug 3 22:53:53 2006 From: jasondunsmore at gmail.com (Jason Dunsmore) Date: Thu, 3 Aug 2006 15:53:53 -0700 Subject: [tbnl-devel] Bug report: no logging with stand-alone SBCL Message-ID: <93f5c6d70608031553j215d243age45ece6544da06bf@mail.gmail.com> I tried setting up logging in TBNL 0.9.10 with stand-alone SBCL 0.9.14 by setting *log-lisp-backtraces-p*, *use-apache-log*, *log-lisp-warnings-p*, and *log-lisp-errors-p* to T and setting (log-file) to "/tmp/tbnl.log", but nothing is logged in the Apache error log file or in /tmp/tbnl.log. After I set *show-lisp-backtraces-p* to T, I am able to see the backtrace in the web browser, but still nothing is logged. When I use this configuration with TBNL and mod_lisp + Apache, backtraces are logged in the Apache error log file. Jason From edi at agharta.de Thu Aug 3 23:23:04 2006 From: edi at agharta.de (Edi Weitz) Date: Fri, 04 Aug 2006 01:23:04 +0200 Subject: [tbnl-devel] Bug report: no logging with stand-alone SBCL In-Reply-To: <93f5c6d70608031553j215d243age45ece6544da06bf@mail.gmail.com> (Jason Dunsmore's message of "Thu, 3 Aug 2006 15:53:53 -0700") References: <93f5c6d70608031553j215d243age45ece6544da06bf@mail.gmail.com> Message-ID: On Thu, 3 Aug 2006 15:53:53 -0700, "Jason Dunsmore" wrote: > I tried setting up logging in TBNL 0.9.10 with stand-alone SBCL > 0.9.14 by setting *log-lisp-backtraces-p*, *use-apache-log*, > *log-lisp-warnings-p*, and *log-lisp-errors-p* to T and setting > (log-file) to "/tmp/tbnl.log", but nothing is logged in the Apache > error log file or in /tmp/tbnl.log. After I set > *show-lisp-backtraces-p* to T, I am able to see the backtrace in the > web browser, but still nothing is logged. If you're using TBNL stand-alone, then you must set *USE-APACHE-LOG* to NIL. If a request comes in from a web browser, i.e. if TBNL can determine that it doesn't come via mod_lisp, it sets the internal value *USE-MODLISP-HEADERS* to NIL for this request. TBNL will log to Apache's log file if /both/ *USE-APACHE-LOG* and *USE-MODLISP-HEADERS* are true. It will log to the log file you've set if *USE-APACHE-LOG* is NIL. Otherwise it won't log at all. See the source code for LOG-MESSAGE. HTH, Edi. From keithorpen at gmail.com Mon Aug 7 16:25:03 2006 From: keithorpen at gmail.com (Keith Orpen) Date: Mon, 7 Aug 2006 12:25:03 -0400 Subject: [tbnl-devel] problems running on sbcl Message-ID: <65b913640608070925j4772b95dk568e357e62b94587@mail.gmail.com> So I finally got tbnl to run smoothly, using lispworks personal. It loads up on SBCL but there is something funny going on with the test pages. I can load one of these pages okay, but every second time i reload it, the browser just hangs there. when i reload it once again, then it works but the test page access count (info.html) has only increased by one, as if the previous request had never happened. I initially thought it was because i was on mac/intel, however tbnl/sbcl had the same problems hosted on linux, and it made no difference moving to the latest sbcl 0.9.15 either. what could be wrong? anyhow, happy that it's running in some form. later, Keith -------------- next part -------------- An HTML attachment was scrubbed... URL: From edi at agharta.de Mon Aug 7 17:16:19 2006 From: edi at agharta.de (Edi Weitz) Date: Mon, 07 Aug 2006 19:16:19 +0200 Subject: [tbnl-devel] problems running on sbcl In-Reply-To: <65b913640608070925j4772b95dk568e357e62b94587@mail.gmail.com> (Keith Orpen's message of "Mon, 7 Aug 2006 12:25:03 -0400") References: <65b913640608070925j4772b95dk568e357e62b94587@mail.gmail.com> Message-ID: On Mon, 7 Aug 2006 12:25:03 -0400, "Keith Orpen" wrote: > So I finally got tbnl to run smoothly, using lispworks personal. It > loads up on SBCL but there is something funny going on with the test > pages. I can load one of these pages okay, but every second time i > reload it, the browser just hangs there. when i reload it once > again, then it works but the test page access count (info.html) has > only increased by one, as if the previous request had never > happened. I initially thought it was because i was on mac/intel, > however tbnl/sbcl had the same problems hosted on linux, and it made > no difference moving to the latest sbcl 0.9.15 either. > > what could be wrong? Is your SBCL built with thread support and are you using a 2.6.x kernel? From keithorpen at gmail.com Tue Aug 8 14:45:08 2006 From: keithorpen at gmail.com (Keith Orpen) Date: Tue, 8 Aug 2006 10:45:08 -0400 Subject: [tbnl-devel] problems running on sbcl In-Reply-To: References: <65b913640608070925j4772b95dk568e357e62b94587@mail.gmail.com> Message-ID: <65b913640608080745u44207a49ib2327126ff02f0df@mail.gmail.com> 2.6 kernel, yes (ubuntu 6.06, if that matters). As for thread support, i thought i had configured this in the sense that i wrote it into the file customize-target-features.lisp: (lambda (features) (flet ((enable (x) (pushnew x features)) (disable (x) (setf features (remove x features)))) ;; Threading support, available on x86/x86-64 Linux only. (enable :sb-thread))) this is as it was given in the INSTALL documentation. However, after I built it, I see that it is not in the file local-target-features.lisp-expr ... should it be? I suspect I have built sbcl without thread support despite my intentions, but I'm not sure how to tell. Keith On 8/7/06, Edi Weitz wrote: > > On Mon, 7 Aug 2006 12:25:03 -0400, "Keith Orpen" > wrote: > > > So I finally got tbnl to run smoothly, using lispworks personal. It > > loads up on SBCL but there is something funny going on with the test > > pages. I can load one of these pages okay, but every second time i > > reload it, the browser just hangs there. when i reload it once > > again, then it works but the test page access count (info.html) has > > only increased by one, as if the previous request had never > > happened. I initially thought it was because i was on mac/intel, > > however tbnl/sbcl had the same problems hosted on linux, and it made > > no difference moving to the latest sbcl 0.9.15 either. > > > > what could be wrong? > > Is your SBCL built with thread support and are you using a 2.6.x > kernel? > -------------- next part -------------- An HTML attachment was scrubbed... URL: From edi at agharta.de Tue Aug 8 15:11:10 2006 From: edi at agharta.de (Edi Weitz) Date: Tue, 08 Aug 2006 17:11:10 +0200 Subject: [tbnl-devel] problems running on sbcl In-Reply-To: <65b913640608080745u44207a49ib2327126ff02f0df@mail.gmail.com> (Keith Orpen's message of "Tue, 8 Aug 2006 10:45:08 -0400") References: <65b913640608070925j4772b95dk568e357e62b94587@mail.gmail.com> <65b913640608080745u44207a49ib2327126ff02f0df@mail.gmail.com> Message-ID: On Tue, 8 Aug 2006 10:45:08 -0400, "Keith Orpen" wrote: > I suspect I have built sbcl without thread support despite my > intentions, but I'm not sure how to tell. I think you'll have to ask the SBCL experts. I usually don't use SBCL. From jsnell at iki.fi Tue Aug 8 21:42:39 2006 From: jsnell at iki.fi (Juho Snellman) Date: Tue, 8 Aug 2006 21:42:39 +0000 (UTC) Subject: [tbnl-devel] Re: problems running on sbcl References: <65b913640608070925j4772b95dk568e357e62b94587@mail.gmail.com> <65b913640608080745u44207a49ib2327126ff02f0df@mail.gmail.com> Message-ID: wrote: > this is as it was given in the INSTALL documentation. However, after I built > it, I see that it is not in the file local-target-features.lisp-expr ... > should it be? I suspect I have built sbcl without thread support despite my > intentions, but I'm not sure how to tell. If (FIND :SB-THREAD *FEATURES*) returns NIL, you don't have thread support. Otherwise you do. -- Juho Snellman From keithorpen at gmail.com Tue Aug 8 23:06:04 2006 From: keithorpen at gmail.com (Keith Orpen) Date: Tue, 8 Aug 2006 19:06:04 -0400 Subject: [tbnl-devel] problems running on sbcl In-Reply-To: References: <65b913640608070925j4772b95dk568e357e62b94587@mail.gmail.com> <65b913640608080745u44207a49ib2327126ff02f0df@mail.gmail.com> Message-ID: <65b913640608081606v3f4a66cdua55a4c2ff3f7ba8e@mail.gmail.com> It turns out I do have threading enabled. I examined *features* and it does contain :sb-thread. So, perhaps some other features are necessary for this to work. In particular, it seems the kernel needs NPTL support (native posix thread library). According to getconf, I have that also. There are some bugs floating around that reference ubuntu and sbcl though; I guess probably it is some low-level problem related to my particular linux config. I will keep looking and in the meantime lispworks will be the alternative. regards, Keith On 8/8/06, Edi Weitz < edi at agharta.de> wrote: > > On Tue, 8 Aug 2006 10:45:08 -0400, "Keith Orpen" > wrote: > > > I suspect I have built sbcl without thread support despite my > > intentions, but I'm not sure how to tell. > > I think you'll have to ask the SBCL experts. I usually don't use > SBCL. > -------------- next part -------------- An HTML attachment was scrubbed... URL: From edi at agharta.de Wed Aug 16 10:21:00 2006 From: edi at agharta.de (Edi Weitz) Date: Wed, 16 Aug 2006 12:21:00 +0200 Subject: [tbnl-devel] [Xristos Kalkanis] tbnl/sbcl problems Message-ID: This was probably supposed to go to the mailing list. -------------- next part -------------- An embedded message was scrubbed... From: Xristos Kalkanis Subject: tbnl/sbcl problems Date: Wed, 16 Aug 2006 04:27:38 +0100 Size: 2791 URL: From edi at agharta.de Wed Aug 16 12:31:35 2006 From: edi at agharta.de (Edi Weitz) Date: Wed, 16 Aug 2006 14:31:35 +0200 Subject: [tbnl-devel] New version 0.9.11 Message-ID: ChangeLog: Version 0.9.11 2006-08-16 Added note about SBCL problems Download: http://weitz.de/files/tbnl.tar.gz This is just a documentation update. No need to download it, if you have it working. Cheers, Edi. From erick at fsl.org.mx Wed Aug 16 16:13:15 2006 From: erick at fsl.org.mx (Erick Ivaan Lopez Carreon) Date: Wed, 16 Aug 2006 11:13:15 -0500 Subject: [tbnl-devel] [Xristos Kalkanis] tbnl/sbcl problems In-Reply-To: References: Message-ID: <1155744795.4054.6.camel@localhost.localdomain> Hello: On Wed, 2006-08-16 at 12:21 +0200, Edi Weitz wrote: > This was probably supposed to go to the mailing list. > > email message attachment > > -------- Forwarded Message -------- > > From: Xristos Kalkanis > > To: keithorpen at gmail.com > > Cc: edi at agharta.de > > Subject: tbnl/sbcl problems > > Date: Wed, 16 Aug 2006 04:27:38 +0100 > > > > On Mon, 7 Aug 2006 12:25:03 -0400, "Keith Orpen" > gmail.com> wrote: > > > > > So I finally got tbnl to run smoothly, using lispworks personal. It > > > loads up on SBCL but there is something funny going on with the test > > > pages. I can load one of these pages okay, but every second time i > > > reload it, the browser just hangs there. when i reload it once > > > again, then it works but the test page access count (info.html) has > > > only increased by one, as if the previous request had never > > > happened. I initially thought it was because i was on mac/intel, > > > however tbnl/sbcl had the same problems hosted on linux, and it made > > > no difference moving to the latest sbcl 0.9.15 either. > > > > > > what could be wrong? I did some tests and found the following things: - When I use the stand-alone tbnl server the bad behavior happens . - When using Apache2 mod_lisp2 to comunicate with sbcl everything works well What do you think? I'm running the latest Debian Sid packages. > > In latest kmrcl, file sockets.lisp, line 118 (function close-active- > > socket > > you need to change #+sbcl (sb-bsd-sockets:socket-close socket) to > > (close socket)). > > After i did the change you say : On: cl-kmrcl: /usr/share/common-lisp/source/kmrcl/sockets.lisp I did: (defun close-active-socket (socket) ;;#+sbcl (sb-bsd-sockets:socket-close socket) #+sbcl (close socket)) #-sbcl (close socket)) Everything works well :) Thank you! -- Erick Ivaan Lopez Carreon -- erick at fsl.org.mx PGP Key 1024D/9741C03A 2004-11-19 Key fingerprint = 1764 3C6F B433 B2DD 9029 98B1 8E6B 58D3 9741 C03A Participa en la FUNDACION DE SOFTWARE LIBRE, A.C. http://www.fsl.org.mx Somos una asociaci?n civil dedicada a promover, difundir y apoyar el uso y desarrollo del Software Libre en M?xico. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part URL: From ccalca at essex.ac.uk Thu Aug 17 03:45:35 2006 From: ccalca at essex.ac.uk (Xristos Kalkanis) Date: Thu, 17 Aug 2006 04:45:35 +0100 Subject: [tbnl-devel] [Xristos Kalkanis] tbnl/sbcl problems Message-ID: <7C4BB735-6C64-4E28-9F06-1E5E495F479A@essex.ac.uk> > I did some tests and found the following things: > > - When I use the stand-alone tbnl server the bad behavior happens . > > - When using Apache2 mod_lisp2 to comunicate with sbcl everything > works > well > > What do you think? > > I assume its because mod_lisp keeps one connection open to the lisp side therefore the actual socket handling in terms of http requests/replies happens in apache. I don't use apache/mod_lisp myself, once i finish my project i'll probably deploy it on lighttpd/fastcgi. From travis at travislists.com Mon Aug 21 15:59:18 2006 From: travis at travislists.com (Travis Cross) Date: Mon, 21 Aug 2006 15:59:18 +0000 Subject: [tbnl-devel] [Xristos Kalkanis] tbnl/sbcl problems In-Reply-To: References: Message-ID: <44E9D856.7080200@travislists.com> Edi Weitz wrote: > In latest kmrcl, file sockets.lisp, line 118 (function close-active- > socket you need to change #+sbcl (sb-bsd-sockets:socket-close > socket) to (close socket)). > > This change was introduced some months ago via some patches for > better sbcl support sent to the mailing list. With this change, TBNL > works ok for me in Linux and OSX under latest sbcl. File uploads used > to work under sbcl 0.9.13 but they no longer do (sbcl 0.9.15, linux & > osx). Interesting. With the above change, all the tests, including file uploads, seem to work for me on linux / sbcl-0.9.15 / kmrcl-1.88 / rfc2388-1.2 / mod_lisp2-1.2. The tests all passed both with and without apache in the picture. Cheers, -- Travis From edi at agharta.de Mon Aug 21 16:47:53 2006 From: edi at agharta.de (Edi Weitz) Date: Mon, 21 Aug 2006 18:47:53 +0200 Subject: [tbnl-devel] [Xristos Kalkanis] tbnl/sbcl problems In-Reply-To: <44E9D856.7080200@travislists.com> (Travis Cross's message of "Mon, 21 Aug 2006 15:59:18 +0000") References: <44E9D856.7080200@travislists.com> Message-ID: On Mon, 21 Aug 2006 15:59:18 +0000, Travis Cross wrote: > Edi Weitz wrote: > > [...] I didn't write the part you quoted. From erik.enge at gmail.com Tue Aug 22 13:32:35 2006 From: erik.enge at gmail.com (Erik Enge) Date: Tue, 22 Aug 2006 09:32:35 -0400 Subject: [tbnl-devel] Suggestion: export GET-TRACEBACK. Message-ID: <58f839b70608220632t167d6aevb00292a4f2083c6e@mail.gmail.com> Hi, I find myself using GET-TRACEBACK. In development mode, I let my unhandled conditions bubble up to the browser window but this would be undesirable for production mode, so for it I simply log that an error happened and display a nice error page. However, I find myself wishing that I had logged the condition using GET-TRACEBACK, like it's done in development mode, for easier debugging. Make sense? Maybe I missing something or this is an internal you're not comfortable exporting. I thought I'd ask anyway. Thanks, Erik. From edi at agharta.de Tue Aug 22 14:39:33 2006 From: edi at agharta.de (Edi Weitz) Date: Tue, 22 Aug 2006 16:39:33 +0200 Subject: [tbnl-devel] Suggestion: export GET-TRACEBACK. In-Reply-To: <58f839b70608220632t167d6aevb00292a4f2083c6e@mail.gmail.com> (Erik Enge's message of "Tue, 22 Aug 2006 09:32:35 -0400") References: <58f839b70608220632t167d6aevb00292a4f2083c6e@mail.gmail.com> Message-ID: On Tue, 22 Aug 2006 09:32:35 -0400, "Erik Enge" wrote: > I find myself using GET-TRACEBACK. In development mode, I let my > unhandled conditions bubble up to the browser window but this would > be undesirable for production mode, so for it I simply log that an > error happened and display a nice error page. However, I find > myself wishing that I had logged the condition using GET-TRACEBACK, > like it's done in development mode, for easier debugging. > > Make sense? Maybe I missing something or this is an internal you're > not comfortable exporting. I thought I'd ask anyway. This one doesn't suffice? http://weitz.de/tbnl/#*log-lisp-backtraces-p* Cheers, Edi. From erik.enge at gmail.com Tue Aug 22 14:48:28 2006 From: erik.enge at gmail.com (Erik Enge) Date: Tue, 22 Aug 2006 10:48:28 -0400 Subject: [tbnl-devel] Suggestion: export GET-TRACEBACK. In-Reply-To: References: <58f839b70608220632t167d6aevb00292a4f2083c6e@mail.gmail.com> Message-ID: <58f839b70608220748y4ac8cbbby9e655d595ae8c98e@mail.gmail.com> On 8/22/06, Edi Weitz wrote: > This one doesn't suffice? > > http://weitz.de/tbnl/#*log-lisp-backtraces-p* Not unless I'm misunderstanding. When the application is in production mode the conditions never reach TBNL as I'm handling them by logging the error and returning some HTML (an error page). Here's the code that is invoked when an error occurs: (defun web-when-error (user dbcon request reply condition) (declare (ignore dbcon reply)) (if (eql *sys-class* :dev) (error condition) (progn (tbnl:log-message :error (tbnl::get-backtrace condition)) (case (type-of condition) (db:validation-error (validation-error-page user request condition)) (t (an-error-occurred-page user)))))) Am I missing something embarrassingly obvious? Erik. From edi at agharta.de Tue Aug 22 15:57:55 2006 From: edi at agharta.de (Edi Weitz) Date: Tue, 22 Aug 2006 17:57:55 +0200 Subject: [tbnl-devel] Suggestion: export GET-TRACEBACK. In-Reply-To: <58f839b70608220748y4ac8cbbby9e655d595ae8c98e@mail.gmail.com> (Erik Enge's message of "Tue, 22 Aug 2006 10:48:28 -0400") References: <58f839b70608220632t167d6aevb00292a4f2083c6e@mail.gmail.com> <58f839b70608220748y4ac8cbbby9e655d595ae8c98e@mail.gmail.com> Message-ID: On Tue, 22 Aug 2006 10:48:28 -0400, "Erik Enge" wrote: > Here's the code that is invoked when an error occurs: > > (defun web-when-error (user dbcon request reply condition) > (declare (ignore dbcon reply)) > (if (eql *sys-class* :dev) > (error condition) > (progn > (tbnl:log-message :error (tbnl::get-backtrace condition)) > (case (type-of condition) > (db:validation-error > (validation-error-page user request condition)) > (t (an-error-occurred-page user)))))) > > Am I missing something embarrassingly obvious? No, it looks like I just missed something in your first email. I think it is possible to achieve what you have now with a combination of *HTTP-ERROR-HANDLER*, AUX-REQUEST-VALUE, and *LOG-LISP-BACKTRACES-P*, but it'd certainly be kind of tricky. I'll just export GET-BACKTRACE in the next release. Cheers, Edi. From edi at agharta.de Wed Aug 23 11:13:02 2006 From: edi at agharta.de (Edi Weitz) Date: Wed, 23 Aug 2006 13:13:02 +0200 Subject: [tbnl-devel] New Hunchentoot release 0.1.5 Message-ID: ChangeLog: Version 0.1.5 2006-08-23 Connection headers are separated by commas, not semicolons Version 0.1.4 2006-08-22 Refactored streams.lisp to appease LW compiler (thanks to Martin Simmons) Changed handling of version string Changed package handling in system definition (thanks to Christophe Rhodes) Download: http://weitz.de/files/hunchentoot.tar.gz Cheers, Edi. From edi at agharta.de Mon Aug 28 08:58:49 2006 From: edi at agharta.de (Edi Weitz) Date: Mon, 28 Aug 2006 10:58:49 +0200 Subject: [tbnl-devel] New releases TBNL 0.10.0 / Hunchentoot 0.2.0 Message-ID: TBNL: ----- ChangeLog: Version 0.10.0 2006-08-28 Based LispWorks version of TBNL on Hunchentoot infrastructure Added "easy" handlers Exported GET-BACKTRACE (suggested by Erik Enge) Download: http://weitz.de/files/tbnl.tar.gz Hunchentoot: ------------ ChangeLog: Version 0.2.0 2006-08-28 Serves as infrastructure for TBNL now (to replace KMRCL) For HTTP/1.1 only send 'Keep-Alive' headers if explicitely requested Download: http://weitz.de/files/hunchentoot.tar.gz LispWorks users please note that Hunchentoot is now a mandatory installation if you want to use TBNL - it replaces KMRCL. You can still use TBNL as before, but the installation procedure has changed a bit - see documentation. For other Lisps the installation hasn't changed. Please note that this infrastructure change and the new "easy" handlers should be considered experimental. Please report problems to the mailing list. Thanks. Have fun, Edi. From edi at agharta.de Wed Aug 30 22:35:13 2006 From: edi at agharta.de (Edi Weitz) Date: Thu, 31 Aug 2006 00:35:13 +0200 Subject: [tbnl-devel] New versions 0.10.1 (TBNL) and 0.2.2 (Hunchentoot) Message-ID: TBNL: ----- ChangeLog: Version 0.10.1 2006-08-31 Only LispWorks: Set read timeout to NIL if connected to mod_lisp Download: http://weitz.de/files/tbnl.tar.gz Hunchentoot: ------------ ChangeLog: Version 0.2.2 2006-08-31 Skip START-OUTPUT advice completely if working for TBNL Version 0.2.1 2006-08-28 Added write timeouts for LW 5.0 Updated LW links in documentation Download: http://weitz.de/files/hunchentoot.tar.gz Cheers, Edi.