Hello welcome back to another update of a racist, and we got a few things this week done. One of them is the um output formatting for code quality reports, so here's an example of a um merge request and there's still one block: that's related to God quality reports itself, which I want to spend some time on. I already asked for help and the ticket has been created. That is this part. Now it says no changes to code quality. It often says filter loads, but ideally I've seen this happen.

It will just State what has changed um regardless. Let's go to the pipeline because the pipeline failed because I introduced a change that is uh not permitted and we now have the code quality report here available. So if I click on it, we can find all of the code quality issues. In this case they are two one is a blocker and one is info. This will not trigger a field pipeline, but one that is more concerning is verify that build arcs, don't contain sensitive information.

In this case it's API token, which shouldn't be there probably now. The other thing is that if we go back to the merge request and we go to changes, it should load here, but that's the weird part I've seen it do this often, but for some reason sometimes it doesn't. There seems to be a um bug somewhere and now it thinks that there is no report. I, don't know, maybe I'm doing something wrong.

Let me know so: I've was able to scrape a bunch of Docker files from GitHub and I just searched for python from python tree, and that gave like 80 000 results, and that seems great. But okay Department allows the first one on the pages, so I create a script that would get the docker files and store them locally and I did that for python, 3.6, seven, eight and nine and ten. However, um you would expect and to have four thousand or five thousand files, but I only got like 1200.

Which is far from enough to do any meaningful training. um So if anyone has an idea on how I can get way more Docker files, maybe we can access them directly from uh gitly uh I, don't know otherwise. I'm gonna have to find an alternative.

I was thinking about reverse building, Docker files or looking at Docker Hub, seeing what our most common uh most most popular Docker files, Docker images and retrieve the docker file logos, I, don't know we'll see at least I'm gonna be able to test these Docker files and see if doctor actually passes on them and that's also what's the next step for next week is I'm gonna, add a feature to format and Docker file and a feature to correct errors.

So, for example, what I mentioned earlier in the um the um build arcs like this is insecure goat and this is secure code. So we can just easily change that based on what was insecure and what is secure. So I want to add that and format the file, so I can actually format all of the docker files that I'm collecting to be actually representable against each other. So we have normalized data that we can actually use for training.

um That's it for this update, um see you next Friday. Thank you for watching bye,.