509840 - text files seen as binary in directory listings

Status: RESOLVED DUPLICATE of bug 290800

(Core Graveyard :: File Handling, defect)

Description

[Matěj Cepl]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.1.3pre) Gecko/20090811 Shiretoko/3.5.3pre
Build Identifier: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.1.3pre) Gecko/20090811 Shiretoko/3.5.3pre

(originally filed as https://bugzilla.redhat.com/show_bug.cgi?id=494668)

I'm looking at a directory listing of text files, eg
http://kojipkgs.fedoraproject.org/mash/rawhide-20090407/logs/
Every file there is text, but after clicking on a few, firefox
will begin to think that they are binary and offer to download
them rather than display them.  Seems randomish and not related
to any particular name or content.

The key for reproduction is that first time it usually opens the
file as text, but when you go back and the reopen the file, it
decides it is actually binary.

Able to reproduce with Fedora build as well as with this Mozilla 3.5.3pre binary.

Reproducible: Always

Steps to Reproduce:
1.see above
2.
3.

Comment 1

[Matěj Cepl]

actually it might be somehow related to bug 290800 ... these files have some unprintable characters in them (Ctrl-H, Ctrl-M in nonstandard places, etc).

Comment 2

[Matthias Versen [:Matti]]

I confirmed (with a SM with the old download dialog) that this is caused by content sniffing that Gecko is only using for Apache servers with text/plain and this can only be caused by binary chars in the first x bytes of the document.

I will confirm this bug because this should not be inconsistent, it seems to detect this only if the file is on the cache.

Comment 3

[Michal 'hramrach' Suchanek]

There is already bug 247509 and Bug 266410 about this

Comment 4

[Matthias Versen [:Matti]]

marking as dupe of bug 290800 because it's the same issue