From nobody@hyperreal.com Tue Jan 7 15:03:22 1997 Received: by taz.hyperreal.com (8.8.3/V2.0) id PAA22764; Tue, 7 Jan 1997 15:03:22 -0800 (PST) Message-Id: <199701072303.PAA22764@taz.hyperreal.com> Date: Tue, 7 Jan 1997 15:03:22 -0800 (PST) From: Tino Wildenhain Reply-To: wildenht@komm.rz.ph-erfurt.de To: apbugs@hyperreal.com Subject: LanguagePriority seems to not work, and Accept-Language: fo,bar,* with * at end also not. X-Send-Pr-Version: 3.2 >Number: 94 >Category: mod_negotiation >Synopsis: LanguagePriority seems to not work, and Accept-Language: fo,bar,* with * at end also not. >Confidential: no >Severity: non-critical >Priority: medium >Responsible: apache >State: closed >Class: sw-bug >Submitter-Id: apache >Arrival-Date: Tue Jan 7 15:10:01 1997 >Last-Modified: Sat Jun 28 13:48:50 PDT 1997 >Originator: wildenht@komm.rz.ph-erfurt.de >Organization: >Release: 1.2b3 >Environment: AIX 4.1 C for AIX (IBM) >Description: 1. I've set up Apache to support Content-Negotiation by setting the apropriate AddLanguage's, LanguagePriority, in srm.conf, and MultiViews for all Dirs. LanguagePriority is set only to "de" (setting to "de en fr" produces the same effect) 2. If an browser not capable to send Accept-Language: (also tested with telnet) mainly the English page apears, if not available any other language, eccept the expected "german". 3. If the Browser sends, p.e. Accept-Language: de,en,fr it gets the german page, as expected. But if the Browser sends Accept-Language: de,en,fr,* (with trailing asterisk), it gets the englisch page in most cases >How-To-Repeat: An URL to test is http://www.ph-erfurt.de/other/ there should apear a "Welcome" in many languages, depending on language preferences. >Fix: none until now. still trying >Audit-Trail: State-Changed-From-To: open-feedback State-Changed-By: marc State-Changed-When: Thu Feb 6 22:43:49 PST 1997 State-Changed-Why: Should be fixed in revision 1.31 of mod_negotiation.c, which will be available in the next beta. A patch is available at "http://www.apache.org/dist/patches/apply_to_1.2b6/", please let me know if it fixes your problem. From: Marc Slemko To: apbugs@apache.org Subject: mod_negotiation/94 Date: Sun, 6 Apr 1997 15:41:44 -0600 (MDT) ---------- Forwarded message ---------- Date: Mon, 24 Feb 1997 03:47:57 +0100 From: Tino Wildenhain To: marc@hyperreal.com Subject: Re: Changed information for PR mod_negotiation/94 Am 07-Feb-97 schrieb marc@hyperreal.com: >Synopsis: LanguagePriority seems to not work, and Accept-Language: fo,bar,* >with * at end also not. >State-Changed-From-To: open-feedback >State-Changed-By: marc >State-Changed-When: Thu Feb 6 22:43:49 PST 1997 >State-Changed-Why: >Should be fixed in revision 1.31 of mod_negotiation.c, >which will be available in the next beta. >A patch is available at >"http://www.apache.org/dist/patches/apply_to_1.2b6/", >please let me know if it fixes your problem. Hi Marc, first the good message: it fixes a little bit: No Accept: -> browser gets the default Accept: de browser gets german Accept: de,en,* browser gets german if possible or en, or ... Accept: de,* browser gets englisch hu? schould not be, what? Current Configuration: LanguagePriority de en fr but now we can live with it. But the mod_negotiation seems to have another problem: the vary-maps dont work at all, I ve set up a test with a file picture.gif (62079 bytes) and a file picture.jpg (22578 bytes) now one of the tests: HEAD /other/picture.var HTTP/1.0 Accept: image/gif;q=0.5;maxbytes=30000, image/jpeg;q=0.4, */* HTTP/1.1 200 OK Date: Mon, 24 Feb 1997 04:42:01 GMT Server: Apache/1.2b7 Connection: close Content-Type: image/gif Expires: Mon, 24 Feb 1997 04:42:01 GMT Last-Modified: Sat, 11 Jan 1997 16:49:31 GMT ETag: "9864-f27f-32d7c49b" Content-Length: 62079 Accept-Ranges: bytes but I expected the picture.jpg here, because the file as .gif is 62079 bytes and therefore larger than maxbytes. have you tried this? regards, -- *-----------------------------------------------------* | Tino Wildenhain | | Nordhäuser Str. 63 WH I/113, 99089 Erfurt, Germany | | Tel: +49 361 737 1878 | | http://www.home.pages.de/~cdTv/ | | mailto:wildenht@komm.rz.ph-erfurt.de | *-----------------------------------------------------* Severity-Changed-From-To: serious-non-critical Severity-Changed-By: pcs Severity-Changed-When: Fri May 30 05:55:06 PDT 1997 From: Paul Sutton To: wildenht@komm.rz.ph-erfurt.de Subject: Re: mod_negotiation/94 Date: Fri, 30 May 1997 14:06:43 +0100 (BST) On Sun, 6 Apr 1997, Marc Slemko wrote: > No Accept: -> browser gets the default > Accept: de browser gets german > Accept: de,en,* browser gets german if possible or en, or ... > Accept: de,* browser gets englisch hu? schould not be, what? > > Current Configuration: > LanguagePriority de en fr > but now we can live with it. This is an old bug report, but still open in the bug database. The LanguagePriority header has no effect if the request includes an Accept-Language: header. This is because under HTTP/1.1 Accept-Language can include priority values (q=X) for each language. It would be difficult to sensibly merge the explicity q values with implicity LanguagePriority priorities. > the vary-maps dont work at all, I ve set up a test with > a file picture.gif (62079 bytes) and a file picture.jpg (22578 bytes) Do you have the file size listed in the type map itself? If a type map is used, Apache will not get any information from the file itself(*), so put the size into the type map (using Content-Length: lines). Also HTTP/1.1 no longer defines the maxbytes parameter, so do not expect browsers to send it. However Apache should still honor it if it is present. If your type-map does include sizes and either the maxbytes is being ignored, or the smallest most-acceptable variant is not being selected, send a copy of your type-map so we can have a look at the problem. (*) Actually Apache will use the information from the file when it returns the file in the response, which can lead to inconsistency, but this is a whole different problem. Paul pcs@apache.org From: Tino Wildenhain To: Paul Sutton Subject: Re: mod_negotiation/94 Date: Fri, 30 May 1997 15:56:17 +0100 Am 30-Mai-97 schrieb Paul Sutton: >On Sun, 6 Apr 1997, Marc Slemko wrote: >> No Accept: -> browser gets the default >> Accept: de browser gets german >> Accept: de,en,* browser gets german if possible or en, or ... >> Accept: de,* browser gets englisch hu? schould not be, what? >> >> Current Configuration: >> LanguagePriority de en fr >> but now we can live with it. >This is an old bug report, but still open in the bug database. >The LanguagePriority header has no effect if the request includes an >Accept-Language: header. This is because under HTTP/1.1 Accept-Language >can include priority values (q=X) for each language. It would be difficult >to sensibly merge the explicity q values with implicity LanguagePriority >priorities. but the server should correctly handle the * value, and give the next matching language back. The current behavior leeds into confusion by the user, if he has the browser configured this way: de,en,* (or includet the * in any way) >> the vary-maps dont work at all, I ve set up a test with >> a file picture.gif (62079 bytes) and a file picture.jpg (22578 bytes) >Do you have the file size listed in the type map itself? If a type map is >used, Apache will not get any information from the file itself(*), so put >the size into the type map (using Content-Length: lines). >Also HTTP/1.1 no longer defines the maxbytes parameter, so do not expect >browsers to send it. However Apache should still honor it if it is >present. If your type-map does include sizes and either the maxbytes is >being ignored, or the smallest most-acceptable variant is not being >selected, send a copy of your type-map so we can have a look at the >problem. Ok, here the file: URI: pce500 URI: pce500.jpg Content-type: image/jpeg; qs=0.1 Content-lenght: 22578 URI: pce500.gif Content-type: image/gif; qs=0.1 Content-Length: 62079 the File can be found at http://www.ph-erfurt.de/other/pce500.var >(*) Actually Apache will use the information from the file when it returns >the file in the response, which can lead to inconsistency, but this is a >whole different problem. regards, -- +-----------------------------------------------------+ | Tino Wildenhain | | Nordhäuser Str. 63 WH I/113, 99089 Erfurt, Germany | | Tel: +49 361 737 1878 | | http://www.home.pages.de/~cdTv/ | | mailto:wildenht@komm.rz.ph-erfurt.de | +-----------------------------------------------------+ State-Changed-From-To: feedback-closed State-Changed-By: dgaudet State-Changed-When: Sat Jun 28 13:48:50 PDT 1997 State-Changed-Why: A fix has been committed to 1.3-dev and 1.2.1-dev. Dean >Unformatted: