« first day (64 days earlier)      last day (2841 days later) » 

12:12 AM
@LuisMendo Oh right that's what I meant. I just mis-typed
 
 
1 hour later…
1:23 AM
@LuisMendo I have the web hook written and installed. If you could set that up within the MATL repo settings that would be great. Here is a screen shot of the settings (minus the secret obviously)
And then select the "Releases" option so that we only refresh on changes to the releases.
If you send me an email at suever@gmail.com, I can email you the secret key to paste in there.
 
 
14 hours later…
3:02 PM
@Suever Done! How do I know if it was successful?
 
3:21 PM
@LuisMendo Should show a little green checkmark next to it
 
@Suever Great, then it has worked
 
Excellent!
We'll have to ensure everything is great when you create the next release but it should be working great.
Oh and if you "redo" a release like you did with 18.3.0, the updates will be fetched as well so you don't have to worry about that
 
3:37 PM
@Suever Wow! Not that plan I on doing it, but it's good that even that works!
 
Takes away some of the gravity of a new release
 
@Suever Is there a limit as to how many releases can be stored on the server? Should I take care to issue releases sparingly?
 
Nope. So a release doesn't really do much except puts an entry in a database saying that a release exists. I don't actually download the source code until it is used
And even then a release is only a few MB so it's not a big deal at all
If it becomes a problem I can always remove old unused releases after a period of time and re-download on demand if someone needs them for an answer
basically at any time I can blow away the entire folder holding all the source code and it will be recreated on demand
I'll be open sourcing the whole project soon so I can show you how all of that works
 
@Suever Oh, great. That's a very good approach. So it's much like a cache
 
Exactly
So a new release updates the database and clears the cache for a particular release if it's different
So very little overhead for a release
 
 
5 hours later…
8:57 PM
I just committed this. It may be useful in image challenges:
In colormap, imshow and imwrite, colormaps with some entry greater than 1 are normalized by converting to uint8, then to double, and then dividing by 255.
This will often avoid (explicitly) dividing by 255 to make the colormap entries lie between 0 and 1
 
9:27 PM
@LuisMendo Oh yea that would be pretty handy for sure.
 

« first day (64 days earlier)      last day (2841 days later) »