Re: [fossil-users] Fossil CGI problem
Any reason for not using one of the established web servers that actually implement the specs correctly? E.g. with nginx you can either hook up fossil http via inetd or directly proxying fossil server. If I understand it correctly fossil uses non-parsed headers in its answers (runs as a non-pased header script). In the CGI 1.1 specification you can find this: 5 5.1 NPH Scripts Identification The server MAY support NPH (Non-Parsed Header) scripts; these are scripts to which the server passes all responsibility for response processing. Bauk didn't imlpement it - if I got that right - but that's right because the specs are implemented correctly. I like Bauk and its features and won't install other servers/applications. Regards, Feri ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
2011/1/17 Kulcsár Ferenc crusa...@netbsd.hu If I understand it correctly fossil uses non-parsed headers in its answers (runs as a non-pased header script). I don't think so. The webserver on the http://www.fossil-scm.org/ website does not support non-parsed headers but it supports Fossil just fine. I suppose the previous sentence depends on your definition of non-parsed headers. The webserver at http://www.fossil-scm.org/ certainly does look at and parse the result it receives from CGI prior to relaying it back to the requester. -- D. Richard Hipp d...@sqlite.org ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
On Mon, Jan 17, 2011 at 09:32:22AM +0100, Kulcsár Ferenc wrote: Any reason for not using one of the established web servers that actually implement the specs correctly? E.g. with nginx you can either hook up fossil http via inetd or directly proxying fossil server. If I understand it correctly fossil uses non-parsed headers in its answers (runs as a non-pased header script). In the CGI 1.1 specification you can find this: No. Fossil depends on the server implementing 6.3 of RFC 3875 faithfully. Bauk adds a header that the CGI doesn't support (which is bogus for POST requests) and it drops a header that the CGI created (content-length). The latter is actually a big no-no for any web server. Joerg ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
On Fri, Jan 14, 2011 at 12:22:51PM +0100, Kulcsár Ferenc wrote: I'm trying to get fossil setup on a Linux box with Bauk webserver. Any reason for not using one of the established web servers that actually implement the specs correctly? E.g. with nginx you can either hook up fossil http via inetd or directly proxying fossil server. Joerg ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
For the people potentially checking out the Twisted script -- the ticket with the bug just got fixed upstream, so there's basically a one-release window where it was broken (10.2.0). cheers lvh ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
I got in touch with the Bauk's developers. It is a long forum thread about this problem here: http://bauk.ws/forum.jsx?a=browsetid=2 Can you help me with this? TIA, feri ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
On Fri, Jan 14, 2011 at 12:22:51PM +0100, Kulcsár Ferenc wrote: Hello, I'm trying to get fossil setup on a Linux box with Bauk webserver. It's partly successful. I can point to my repo and administer it. The checkout is failing: $ fossil clone http://archlap.aneder.hu/fossil/proba1 proba1.fossil --httptrace Bytes Cards Artifacts Deltas Sent: 53 1 0 0 fossil: server did not reply There are at least two issues with the behavior of the server. First, it is adding Accept-Ranges, even though the CGI didn't say so. As the original request is a POST, this is just a bug in the server. Second and more important, it seems to drop the content-length header from the cgi output. Joerg ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
[fossil-users] Fossil CGI problem
Hello, I'm trying to get fossil setup on a Linux box with Bauk webserver. It's partly successful. I can point to my repo and administer it. The checkout is failing: $ fossil clone http://archlap.aneder.hu/fossil/proba1 proba1.fossil --httptrace Bytes Cards Artifacts Deltas Sent: 53 1 0 0 fossil: server did not reply The debug files are attached. Any ideas? TIA feri HTTP/1.1 200 OK Server: Bauk/2.0.19 Accept-Ranges: bytes Date: Fri, 14 Jan 2011 09:27:46 GMT Connection: close Content-Type: application/x-fossil-uncompressed; charset=utf-8 Cache-Control: no-cache, no-store cfile 6e5087ac955f16aee6b0e84ba5b3d9a331541231 161 151 ¡x Ì90@ÑÞ§ÉÒmÌ8mr(MDã a¬HÁí³ÿÿ ¥½øç°ååµ Ã§ gYª¸B$I÷NOÅUÜ J ã1(4Ù1¢sãÈFuYôÐÕ×8Aûú®óO¶cÿFÜaÌk BëBbk }7Îúè½oEmçÄÖÛ-a clone_seqno 0 push fc6203988173a997794cf732fcbaf0a91a7fdf76 709f51faeb0e3ab71a8640f59b43629813ee8c3a # timestamp 2011-01-14T09:27:46POST /fossil/proba1/xfer/xfer HTTP/1.0 Host: archlap.aneder.hu User-Agent: Fossil/[356e877cea] Content-Type: application/x-fossil-debug Content-Length: 53 clone 3 1 # 8BB3878C636B992E809121954B35AE35D1EE2F03___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
That hostname isn't supposed to be internet-accessible, is it? I've written an admittedly ad-hoc Twisted script which sort-of works (depending on versions, you might get a duplicate Content-Type, which, depending on browser (ie Chrome), may mean the CSS doesn't work). If you want, you can try that, perhaps we can compare responses to see what goes wrong. cheers lvh ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
I do wonder why there aren't any newlines. Was that lost in the paste, or is that really what Bauk produces as a response? If my spec-memory is correct, HTTP defines a newline as a CR LF, there's supposed to be one after each header, and the request ends with a double newline. I can't see that in the paste. It's possible, that Fossil, like Chrome is less forgiving of bogus input? (I can hardly blame it, Accept and cookie-specific headers are already difficult enough to parse without having to guess where they end...) cheers lvh ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
You can access my host now: http://crusader.dyndns.info/fossil/proba1/index In mcedit each lines in the files are ended with ^M. Thanks. feri ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
Re: [fossil-users] Fossil CGI problem
If I'm reading src/http.c correctly, that basically only happens if iLength remains set to its original value of -1, so the response is probably accurate. My best guess is that it's either a Bauk bug or misconfiguration issue. Unfortunately I've never heard of Bauk, so I can't really help you there. I have added a smaller version of my twisted script[1] which may help you unless you really need to use Bauk. Fair word of warning: maybe be flakey with some clients in latest version of twisted because of a http header regression[2]). Edit lines 17 and 18. Put your fossil repo in repositories/Yourname.fossil, along with a file Yourname.cgi, which just contains: repository: Yourname.fossil (Twisted will already run this with fossil, if fossil is on your PATH.) Run python script.py, and you can access your repo at http://localhost:8080/Yourname (the name it's available under is the first argument to root.putChild on line 18) [1] http://www.bpaste.net/show/12921/ [2] Ticket: http://twistedmatrix.com/trac/ticket/4786 Workaround (may not be necessary depending on client): comment out twisted/web/server.py, line 118 : http://twistedmatrix.com/trac/browser/tags/releases/twisted-10.2.0/twisted/web/server.py#L118 cheers lvh ___ fossil-users mailing list fossil-users@lists.fossil-scm.org http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
