Adding dependencies. This should also remove the "dirty hack" introduced by bug 52441
Not happening any time too soon... too much arch work to be done first.
I can't believe this bug is still in there. This means that mimetype handling is broken, and is broken in a way which reminds the user that it's broken *every time* they click on a link which uses a non-preconfigured helper app. This seriously harms the UI of my application, which you can get here - http://bitconjurer.org/BitTorrent/download.html - all I can tell people is that the annoying dialogue you have to click 'yes' on every time is mozilla's fault, and no, there's nothing you can do about it. As far as I can tell apps like xmms are acting as plug-ins rather than helper apps, a non-cross-browser technique they most definitely should not have to do.
If you note, this bug has a nonempty "depends on" field. That field is not there for fun. Right now, there is simply no place to store the full mailcap entry for later retrieval and parsing; arch work needs to be done for that. Since it seems I'm the only one who's interested in actually working on it rather than whining about it, it won't happen till I have time, which will not happen till I finish my thesis and graduate. All that said, it is incredibly unlikely that what you're seeing is caused by _this_ bug unless you depend on being able to execute arbitrary shell commands in mailcap helper entries (rather than just running a program). Does the dialog that comes up get pre-filled with the helper app information correctly? If so, it sounds more likely that your users are just not checking the "never ask me again" checkbox that dialog.
I just tried it with 0.9.8, and I'm seeing considerably better behavior - previously unchecking the 'don't ask again' box didn't work, it kept re-asking anyway (which, you're correct, doesn't appear to be _this_ bug, but I'm having trouble following the maze of bugs related to mimetypes) I consider asking that question even once to be broken behavior - it's already configured, isn't it? That should be pretty clear. But I don't expect to be able to make any headway arguing against that behavior because it isn't broken enough to be something which might get fixed. I also can't figure out if there's an RFE to have /etc/mailcap get re-parsed if an unknown mimetype is encountered, to avoid having to restart mozilla in such cases, but maybe that's hoping for too much...
> I consider asking that question even once to be broken behavior - it's already > configured, isn't it? It is. Unfortunately, the default configuration is often crap. For example, on RedHat Linux up through 7.2 the default "helper" for postscript is lpr. They didn't change it to gv till I filed a bug on them to do it, which was about 3 months ago. So we give the user a chance to stamp their approval on the default config or select something else to use. > I also can't figure out if there's an RFE to have /etc/mailcap get re-parsed There is not, but then again it's currently reparsed any time a type is unknown. We used to cache the result once we found a match, but the cache had major issues and has been turned off (that may have been post-0.9.8). It will not be turned on again without addressing the problem you mention.
The changes to enable "useSystemDefault" that are being made in bug 86640 will make this possible even if bug 78919 is fixed.
not happening in the foreseeable future
Am I mistaken if I conclude this is a security issue? Say one, as could be the default, has an /etc/mailcap entry application/postscript ; gv -safer %s Then with the current partial implementation the effect would be to use gv, but without the -safer flag, right?
That's correct, and a really good point... There may well be other apps which have a "default" mode and a "secure" mode and we could be launching the less secure mode...
*** Bug 186304 has been marked as a duplicate of this bug. ***
Uh, that bug is a security bug. Can we open it if it is indeed a dupe?
I said it should be opened in that bug a few weeks back... but in typical fashion that didn't help any.
Has this bug stagnated? I don't see why it's so hard to do it - I can try to provide a patch if desired... This is a really critical thing to get fixed, IMO.
(In reply to comment #16) > Has this bug stagnated? I don't see why it's so hard to do it - I can try to > provide a patch if desired... > > This is a really critical thing to get fixed, IMO. Agree. Great if you can fix it.
This stagnated because I'm frankly not interested in doing this at this point... If you post a patch, I'd be happy to review and help it land; let me know if you need pointers to the code in Mozilla that handles mailcap files.
nice to see that in 2010, with firefox 4.0 this is still an issue I work on LTSP on linux, we run gnome on the server and firefox on a thin client and to be able to integrate correctly firefox and the gnome desktop, i can add some entry in /etc/mailcap to be able to launch the applications on the server, here is on of mailcap entry that works with chromium but doesn't in firefox. Do you think this bug will be solved soon ? of shall i just wait another 10 year to see if the status will change to "OPEN"? == mailcap entry == application/vnd.ms-excel; /usr/bin/ltsp-remoteapps soffice -no-oosplash -calc '%s'; edit=/usr/bin/ltsp-remoteapps soffice -no-oosplash -calc '%s'; test=test -n "$DISPLAY"; description="Microsoft Excel Document"; nametemplate=%s.xls ===
Mark, see comment 18?
would reopen with a plan and patch