httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexei Kosut <>
Subject Re: Apache's byteranges support
Date Tue, 11 Feb 1997 03:07:02 GMT
On Mon, 10 Feb 1997, Anthony Lam wrote:

> I'm not sure who in your organization is responsible for the HTTP byteranges
> implementation.  So I'm sending this email to all of you.  Please disgard if
> you are not interested. If you know who is the expert of byteranges, please
> foward this message to him/her.

I'm mainly responsible for byterange support in Apache. Thanks for
contacting us. As it happens, we were just looking into our byterange
support earlier this week, and have solved some of the problems you

> I found out two problems of the Apache 1.2b6 server doing byteranges support:
> 1. the multipart delimiter that I query from Apache server is
>         something like --"[3XRGJK45S9KLUF]" (wrong)
>    but later on when Apache serve the data chunks over, the delimiter
> seperate the chunks are --[3XRGJK45S9KLUF] (correct)
> So somehow the "" was mistakenly include in the first case.  Since the
> delimiters don't match, the client (Navigator or Internet Explorer) cannot
> parse the data correctly.

As far as I'm aware, Apache only quotes the string in the
Content-Type header, not in the multipart body itself (if this is
not true, and you could give me an actual example where it does not,
that'd be much appreciated): quoting the delimiter here is perfectly
legal according to RFC 2046 (MIME Part Two: Media Types), and section
5.1.1 even strong reccomends putting boundary strings in quotes.

If a browser is known to misbehave because of this, I will strongly
consider removing the quotes. However, Navigator works fine (I have
not yet had the opportunity to examine IE); see below.

> 2. The header info. from Apache server is using the syntax
> multipart/byteranges, which is stated in the latest HTTP 1.1 spec.  But
> unfortunately, both Netscape 3.0 & Internet Explorer 3.0 are following the
> draft spec multipart/x-byteranges, as well as other servers e.g IIS 2.0,
> Enterprise, FastTrack etc....
> We can certainly ask Microsoft & Netscape to support both syntax in their
> future releases.  But I would like to advise you to use the old syntax
> multipart/x-byteranges so that all the Netscape & IE 3.0 users out there can
> view PDF files on Apache servers.

We have already been in contact with Netscape, and they have told us
that Navigator 4.0b2 will correctly support multipart/byteranges. As I
said earlier, I have not had a chance yet to look at Internet

As for back-compatibility with the current crop of browsers, version
1.2b7 of Apache will contain a workaround for that: we discovered that
Navigator sends a Request-Range header as well as a Range header,
complying with the second Luotonen/Franks internet draft (I believe -
it has expired, so I can't check correctly). So Apache 1.2b7 will,
upon receipt of a Request-Range header, send multipart/x-byteranges
instead of multipart/byteranges. This makes Navigator work correctly;
we tested this thoroughly. Again, I have not had an opportunity to
check Internet Explorer.

1.2b7 should be released shortly, although it is being held up by some
problems unrelated to byteranges; the patch to 1.2b6 is, however,
available at
should you wish to try it yourself.

I should mention, however, that we will probably not make any changes
to Apache that make it uncompliant with HTTP/1.1 (even for HTTP/1.0
requests); making all byterange responses deliver
multipart/x-byteranges would be included in this. That being said, we
are committed to making Apache work with as many browsers as

> I hope you can put the fixes in your next release since Apache is one of the
> most popular servers out there, and we've got lots of compliants about PDF
> and Acrobat 3.0 not working with Apache.

As have we, which is why we decided to look into this. Hopefully, we
can get them all resolved.


Alexei Kosut <>      The Apache HTTP Server

View raw message