Closed danielzgtg closed 1 year ago
Thank you for the effort, but I dislike python and will not add that dep. I feel bad to reject your PRs like this, maybe you could ask beforehand to avoid wasted work?
It's ok. Python is a prototyping language and the average Python script is deleted after just 1 use.
ask [...] wasted
I am trying to get Navi23 supported added quickly. It's more like I wrote the script as a byproduct of following https://github.com/clbr/radeontop/pull/144#issuecomment-1304575948 saying they are imported automatically. Another project I'm working on also includes parsing so I had the ideas fresh in my mind already.
I was not confident in importing all those 26 IDs manually. Scripts don't make mistakes so I use them. Nobody except a computer would have noticed the "OLAND" part. I've written Javascript to parse Wikipedia that took 2+ hours only for me to delete it right away and never publish anything.
you could ask beforehand
I would have asked beforehand if I had to write awk
or sed
. It took me less than an hour to write this. I never cared about this pull request because it got the other pull request ready earlier. On the other hand it would probably take me 40+ reluctant hours to learn awk
properly, and 20 hours to learn anything beyond basic sed
. So basically, Bash/Python = I don't care about wasted time, awk
/sed
= I would care about wasted time.
The old
getamdgpuids.sh
doesn't work anymore on kernel 6.1.cpp
complains aboutdrm/
includes thenasm/
includes and even after those-I
are added it still complains about something in the compiler file. Even when the module is built by a normal kernel, it the "PCIAN" and "CHIP" patterns seem to have been changed away. This causes the old script to only keep adding nonsense likeCHIPSET(,,)
.I ported the getamdgpuids script to parse the C file instead. I included a lot of error checking in case the struct changes again. It works with Python3.10 on https://github.com/torvalds/linux/tree/2f5065a0bc9d0afb45f2ae284bd525d55ce82c33 .