No too far this time. My dad & I are driving about 2 hours to boat/fish/camp on a local river
should be back online by next Thursday
bhuddah
usually the distance is not as important as just being somewhere else.
have a lot of fun.
knnniggett
exactly
bhuddah
:)
knnniggett
On another note, my wife bought a learning German book for a future trip. :-)
bhuddah
yesterday we recorded great footage of a pickpocket here using zm ^^
ah :)
tell her: "viel Erfolg."
knnniggett
It's always nice to hear ZM success stories
bhuddah
still it's a nuissance to have those guys in the cafe at all -.-
knnniggett
will do!
asker_ joined the channel
asker_
morning
knnniggett
morning
bhuddah
howdy.
asker_
so knnniggett: do you think all installs of ZM allow for direct access to event images without auth?
knnniggett
Only if you have authentication turned off
in the options
asker_
no, even if its on, you can directly access
that is what surprised me
knnniggett
when you supply the username & password in the url
asker_
you can construct a URL to directly access an event image
nope
it skips ZM
knnniggett
well I didn't know that to be the case
asker_
it goes directly via apache
so basically, there is no configuration or commentary in ZM anywhere that tells users to stop direct access to the events directory (which is inside www, symlinked)
apparently, the timeline code relies on it too
it directly picks stuff using HTTP URLs from events
bhuddah
some basic auth would probably help there ^^
knnniggett
can you provide details. I cannot duplicate this claim at the moment
I faced this problem when doing basic auth for APIs
chrome was not allowing me to stuff in user@pass
it would repeatedly prompt me with a dialog box
knnniggett
it is user:pass@hostname
wow that didn't work
asker_
yup
worked for me on Safari (when I did basic auth), but like I said in Chrome, it would not respect my usr:pass stuffed in
so I removed basic auth from the API
knnniggett
Well, I guess the best solution is going to be something more than a quick modification
not sure what that is at the moment
asker_
yup, on the positive side however, directly accessing images made scrubbing possible on the mobile app, which is good
routing it via ZMS and doing scrubbing is impossible (unless ZM is running on a powerful system)
rexd666
knnniggett: The events folder being open access I was referring to yesterday when talking about hosts file in misc and you gave the redhat example. It talks about that event folder issue, and used to disable the webgui by default. https://github.com/ZoneMinder/ZoneMinder/blob/m...
Yeah, I remember discussion about this a while ago, thought we merged a few changes that fixed all security issues with directly accessing the jpegs.... so apparently I mis-remembered.
ic0n
Good morning all. rexd666 I'm working on making h264 passthrough work for rtsp streams.
rexd666
Yes really should kill off symlinked dirs in web folder, some overhead for managing access through code, but better. I don’t spend much time on jpegs these days, but the same thing applies to my h264 files except I have event file location in db
ic0n OK i was just going to copy captureandrecord across once I had it sorted in ffmpeg_camera as they are almost identical
Post merge I would probably bring capture and captureandrecord together
ic0n
Maybe we should be thinking about code structure a bit and try to remove duplication. Maybe even use inheritance.
rexd666
lol snap
I was just going with keep duplication for now so no effect on pre-passthrough code, but recognise it is not good, and stupid big functions
ic0n
yeah that's fine... just thinking that if we put the stuff in ffmpeg into the camera base class, then the others can still have their overriden functions just returning zero, but then we don't have to copy over code into any camera type that we are playing with.
I also want to work on local cameras providing h264.
rexd666
I could never see a way to get the stream from libvlc, which is a pain, guess it means I will have to spend some time on ffmpeg smearing issue
I was just thinking I should move my packet buffer somewhere shared, maybe zm_buffer
maybe not as I am using fifo from ffmpeg so maybe zm_ffmpeg
ic0n
I don't see any problems sticking FIFO Buffer code into zm_buffer. Just a different type of buffer. Might be needed elsewhere.
I run a patch in my tree that has a class called FIfoStream... it is used when creating analysis images and stuff... not sure if it is applicable, because it uses a disk based fifo... (or inode based or whatever0.
yeah, it was written by MitchCapper
anyways, given the option, always try for re-usable components.
I'd like to split the analysis stuff out of the Monitor class into it's own class, like Analyzer...but that's for another day.