Re: octet-stream attachment

From: Daniel Stenberg <>
Date: Thu, 29 Jul 1999 13:21:25 +0200 (MET DST)
Message-ID: <>

On Tue, 20 Jul 1999, Tom von Alten wrote:

Hi crowd! I'm back now, slowly trying to get back in speed and in synch with my environment. Allow me a few days to acclimatize! ;-)

> However, in the archive, it was an "application/octet-stream attachment"
> and given a name ending in ".jpg_" My browser shows this as something
> less interesting than an image.

I can't tell how your mail program could find out it was actually an image file.

Hypermail saves the file with the name it got supplied. If the file name had a ".jpg_" extension it will be used.

When you browse that file with your browser it will probably treat it the way the Content-Type it receives has been defined to behave. Since the web server (probably) only recognizes .jpg as a JPEG picture I trust you'll get some other (default) Content-Type for your .jpg_ picture. Probably application/octet-stream or possibly html/text. Thus your browser will not show it as a picture.

> Unfortunately, I don't have the original message to see exactly what it
> looked like - presumably the JPEG attachment *was* encoded as an octet
> stream. (Different than base64, I presume?)

No, you're mixing encoding and content. base64 is encoding and any content may be coded with base64 (or quoted-printable or 7bit or...). A picture should not be sent as a application/octet-stream. If it was a JPEG picture, it should be sent as a 'image/jpeg'.

> My question is: should hypermail have been able to process this?

No, I can't see how it could have.

IF it had been 'image/jpeg' we COULD make hypermail to automatically set a 'image/jpeg'-extension which then could've made that mail to work as you would've wanted.

IF you somehow configure your web server to send all ".jpg*" file names as 'image/jpeg'. Which your mail program probably did. It is possible you can make your browser do this too. I think there's somewhere where you can tell which suffixes that should do what.

> I will try to preserve another sample message before it's processed into
> the archive, but maybe someone can recognize the issue from these clues.

That would bring some light to this, yes.

             Daniel Stenberg -
   ech`echo xiun|tr nu oc|sed 'sx\([sx]\)\([xoi]\)xo un\2\1 is xg'`ol
Received on Thu 29 Jul 1999 01:22:49 PM GMT

This archive was generated by hypermail 2.3.0 : Sat 13 Mar 2010 03:46:11 AM GMT GMT