Re: single

I know the reason why we have it as an #ifdef. It is because of the 
broken nature of the IE browsers. The behaviour of content-dispostions 
is inconsistant on the Microsoft environment(s). In a nutshell, the IE 
that ships with Windows 98 handles content-disposition, than does the 
default IE that ships with Windows2k, and is different again in the 
version of IE that ships in WindowsXP. To compound the issue some, 
patching the IE browser on the notably broken versions makes IE handle 
content-disposition different yet again. Microsoft has this problem 
documented in a handfull of Knowledge Base articles (hence forth KB) on 
their website. The Problem is IE 4 does them one way, IE 5 did them 
another way, and the patch to fix the broken IE 5 did them yet another 
way (that was still broken), IE5.5 did them different, and the IE 6 did 
them yet another way. The problem started with the last version of IE4 
that introduced a new mshtml.dll rendering engine.

I too have done my testing, and that is why I advised that we make this 
an option to enable or disable back when this feature was first 
introduced as a patch. Here is the deal, on some broken, yet widely used 
browsers, content-disposition will prompt the user to save the mp3 
stream, or open it when it should automatically be passed to the 
application tied to the mime/content-type of the data. Other times I 
have observed a situation where the user agent prompts for the user to 
save/open the stream, and when choosing to open the stream prompts the 
user to save/open the stream again because of the content-type working 
as it should, yet the content-disposition forces the question.

The content-disposition header means different thigns to different 
folks. For some people it means the browser should be forced to prompt 
the user how to dispose of the data (save, or open), to other people it 
is a method to force the presentation of URL's a certain way. I have 
found that Mozilla works the best at this type of stuff, and gives fine 
grained control for this situation. However, IE is is still the dominant 
browser, and might be the broken browser you use to access mod_mp3, or 
an interface to mod_mp3. =(

The best way to solve the issue is to modify the mime-tabels on the 
client side to work correctly.

Bruce M. Walker wrote:

>Quoting Jon disnard <[email protected]>:
>  
>
>>You can also use the HTTP content-disposition header to make it work. 
>>That header works well with content-name, and content-type headers to 
>>work-around the broken IE browser. I think but, I'm not sure, that 
>>somebody
>>    
>>
>
>Yeah, me. ;-)
>
>  
>
>>has already hacked together something like that to use the 
>>content-disposition header on each request. I'd have to take a look at 
>>the configure file to make sure, but I think brian made 
>>content-disposition an on/off feature controlled by a configure arg, or
>>a directive, or whatever.
>>    
>>
>
>No, it's still a compile-time #ifdef ...
>
>   mod_mp3.c: #ifdef CONTENT_DISPOSITION
>
>which you have to know to enable.
>
>I'm afraid I don't understand that feature being an #ifdef that is off by
>default, still.  It works and it fixes many (all?) of these PC browser issues. 
>I use it regularly, yet I continue to see questions from 1st-time mod_mp3 users
>saying, "how come it downloads instead of streaming in my Windows browser?"
>
>I didn't "invent" this method of telling browsers what to do: I observed how
>existing websites (eg mp3.com, besonic.de, acidplanet.com, etc.) do it by using
>curl and 'lynx -dump' to see what comes down the tubes.  Ever wonder why people
>can install IE or NS on Windows, go to these music sites and "it just works
>(tm)"?  Then they try mod_mp3, and it doesn't.
>
>I know, "UNIX does it right".  I'm a UNIX guy for crumbs' sake; but when it
>comes down to it, what percentage of browsers using mod_mp3 in the future will
>be IE compared to all others?  Do the math.
>
>But hey, as long as the code's still there for *me* to enable, guess I can't
>complain.  :-)
>
>-bmw
>_______________________________________________
>Mod_mp3 mailing list
>[email protected]
>http://lists.tangent.org/mailman/listinfo/mod_mp3
>
>  
>