search

area515 / Photonic3D

Control software for resin 3D printers
http://photonic3d.com
GNU General Public License v3.0
133 stars 112 forks source link

Bug : Adding printer from github broken #231

Closed kloknibor closed 8 years ago

kloknibor commented 8 years ago

Hi,

When you try to add an printer from github to photonic3d it will return an server error 400. (400: Invalid request )

was using it from windows and is reproduceable, logs can be found here :

LogBundle(1).zip

using my latest version and also the latest version of Wes by the way!

jmkao commented 8 years ago

Hmm, strange, in the Chrome inspector, it seems like there is a 400 error coming from POSTing to the URL that contains the printer JSON. Trying this out in Postman directly, GETing the URL works fine as expected, but POSTing to it will result in a 400 error. However, I'm not sure what the logical flow of this section of code is since the JS stack trace only goes up to a JQuery call.

image

WesGilster commented 8 years ago

Right off hand, this looks like a CORS issue. I saw a commit a while back for a fix to a CORS issue. Did that get fully resolved?

jmkao commented 8 years ago

If you're referring to da2ce0f53a7bd3c107748f15e236d8ada56e077f, that specifically had to do with CORS restrictions on following 302 redirects for repos renamed to "Photonic3D" instead of "Creation-Workshop-Host" (which only impacted my repo).

But CORS issues show up pretty explicitly in the browser console. This seems to be something different related to the HTTP method used.

WesGilster commented 8 years ago

I quick checked this out by replacing the POST request with a GET and the preflight CORS options request fails. However the POSTman request and a basic browser request work fine.

WesGilster commented 8 years ago

It's possible github tightened their CORS restrictions. I check into this tomorrow.

I'm really second guessing if Github is the best place to be keeping these resources anyway. I've got bigger plans for how all of these resources should be shared with one another. I just don't have all of the design work done to replace it with that yet.

WesGilster commented 8 years ago

It looks like they are now using CORS to ensure people use download urls only from their site: default-src 'none'; style-src 'unsafe-inline'

However for small enough files, apparently we can still snag them through their restful API. Checked in a fix.

jmkao commented 8 years ago

I tested this and it's working now.

Between this and the start.sh fix that I just sent to you in a PR, how close are we to generating a new stable release? Stable's printer settings download is probably also not working.

I tried to cherry-pick just the relevant commits, but it's tricky because for changes to printers.js which has other changes in it that are tied to server changes as well.

kloknibor commented 8 years ago

Feel free to use the photonic3d server to actually pull the data from!

WesGilster commented 8 years ago

Merged and pushed a new production release.

kloknibor commented 8 years ago

This bug is back! I can't seem to download any printers from github and I can't reopen this issue.. Sorry

WesGilster commented 8 years ago

I'll need steps to reproduce this bug as it seems to be working fine for me.

kloknibor commented 8 years ago

https://www.youtube.com/watch?v=X41m_nDrS-g&feature=youtu.be Here you can see it, for some reason the area515 one closes right away. This wasn't the case before... This was actually the first time I encountered this! So windows version is broken I suppose for now... But when I run from the gradlew run command it does run, but you will see the error 400 and not downloading templates

WesGilster commented 8 years ago

In order to determine why it closes immediately you'll need to run start.bat from a command prompt instead of just double clicking on the icon. As you can see, when you just double click, it doesn't tell you anything.

As to the other issue, it seems there is either a bug in the repo that you are using, or you have ill formatted json printers. My bet is bad json, but I can't really tell without logs.

WesGilster commented 8 years ago

Downloading and running Release 1.0.0-beta2 works great for me and adding printers also work fine.

kloknibor commented 8 years ago

image

Not much more info when running from the command prompt

but judging from : log.err.txt

My java set-up is broken. (which is funny because compiling works still fine + I didn't temper with the java settings)

kloknibor commented 8 years ago

So winAVR was the problem. It overwrote all settings in the global %PATH% variable so I added it and it worked again. But I can reproduce the problem with area515 set-up too :

https://www.youtube.com/watch?v=BY5lClDpTkU&feature=youtu.be

Might this be because I'm in the EU? Here are the logs : LogBundle (2).zip.pdf

Let me know if more is needed :)! I will try this from work too tomorrow so we can exclude a firewall at my house ;)!

WesGilster commented 8 years ago

Unfortunately, this isn't a problem on the Java end. It looks like a browser issue. Since I can't reproduce this and the browser does have logs, we are in quite a pickle. You'll need to open the chrome integrated debugger(ctrl+shift+i) and look at the network traffic that looks like this: image

There should be a highlighted red resource that is giving you the 400 message. When you find that, you need to take a look at what the response says.

kloknibor commented 8 years ago

I understood the printer not downloading from github wasn't a java bug, but the CMD not starting was :)

But after reading your comment and taken an shower, I was thinking about your comment and thought, Shit! Not again... It might be the stupid Ad Blocker I use! It won't be the first time I wasted time by finding out where elements went only to find out it was the fault of the adblocker....

So I just tried it in incognito mode with chrome and it works like a charm! Sorry for wasting time! At least now we now the adblocker can act up! I will close this now.

Thanks for your help

WesGilster commented 8 years ago

It sounds like the CMD wasn't a problem with Java either, it was WinAVR manipulating the state of your system without your knowledge. The box that you are testing with sounds like it's in a constant state of flux. Until you stabilize the software on that box, you are really going go through some pain.

Eventually you may run into a legitimate issue that can only be reproduced periodically. If you don't have a stable system and the software on your box is constantly changing, you'll never track down what is going on. As James mentioned on that imaging thread, it's important to track down specifically the one factor of that changed on your image. Thanks again for the testing, I'll keep the adblocker in mind.