I decided to read through some of it's source code this morning. It definitely has at least some bugs because it obviously has race conditions. Admittedly I'm not sure quite how much they would...
I decided to read through some of it's source code this morning. It definitely has at least some bugs because it obviously has race conditions. Admittedly I'm not sure quite how much they would matter in practice, but they're there.
The obvious race conditions are ones I often write myself out of laziness, but I don't call that code bug free. When you write logic like
if file not exist
create file
that the channel creation script is full of you always have a race because the OS doesn't promise it the state of file won't change before you're done assuming it hasn't changed. Normally you'll get lucky though.
This is a neat idea. As it states up front, it is a bit overselling itself with the title, but neat nonetheless. I assume it would eventually hit scaling issues with particularly huge...
This is a neat idea. As it states up front, it is a bit overselling itself with the title, but neat nonetheless.
I assume it would eventually hit scaling issues with particularly huge organizations because I don't think the tools for users or groups are really meant to support millions of entries, but that would probably be fixable somehow. The thing that would turn out to be a huge issue for me is that it doesn't seem like it would be particularly capable of supporting editing or deleting, both of which are extremely important features to me.
I do really like the simplicity of the git integrations and such. The general concept of "you can authenticate for as long as you are capable of opening a valid ssh connection" is pretty cool on that front when compared to the much more bespoke auth mechanisms that would normally be present to do similar actions.
And a nitpick: it bothers me a bit that it calls out the potential attack surface added by the common-ish mechanism of shoving the text through random, usually complicated, HTML rendering engines but then goes on to shove the text through some random python library of unknown complexity and potential attack surface.
The spinning willow leaf strikes the carp in the lower body, tearing the muscle and bruising the guts! The carp stands up! I will be taking the weekend off please don't tell my manager
famous last words
I decided to read through some of it's source code this morning. It definitely has at least some bugs because it obviously has race conditions. Admittedly I'm not sure quite how much they would matter in practice, but they're there.
The obvious race conditions are ones I often write myself out of laziness, but I don't call that code bug free. When you write logic like
that the channel creation script is full of you always have a race because the OS doesn't promise it the state of file won't change before you're done assuming it hasn't changed. Normally you'll get lucky though.
This is a neat idea. As it states up front, it is a bit overselling itself with the title, but neat nonetheless.
I assume it would eventually hit scaling issues with particularly huge organizations because I don't think the tools for users or groups are really meant to support millions of entries, but that would probably be fixable somehow. The thing that would turn out to be a huge issue for me is that it doesn't seem like it would be particularly capable of supporting editing or deleting, both of which are extremely important features to me.
I do really like the simplicity of the git integrations and such. The general concept of "you can authenticate for as long as you are capable of opening a valid ssh connection" is pretty cool on that front when compared to the much more bespoke auth mechanisms that would normally be present to do similar actions.
And a nitpick: it bothers me a bit that it calls out the potential attack surface added by the common-ish mechanism of shoving the text through random, usually complicated, HTML rendering engines but then goes on to shove the text through some random python library of unknown complexity and potential attack surface.
The spinning willow leaf strikes the carp in the lower body, tearing the muscle and bruising the guts!
The carp stands up!
I will be taking the weekend off please don't tell my manager
https://gitlab.com/edouardklein/suc