Per WP 13472 updated h1daqfw0 to use the new auto-reconfiguring frame writer. This is the same code has been running on h1daqfw2 for months in a test mode.
I also did an inplace upgrade of h1daqfw0 to Debian 12, which is what h1daqfw2 is running and what we have tested as.
Erik helped with package wrangling for Debian 12. I had been running h1daqfw2 from the experimental repository (which is a snapshot of the continous integration system). We were able to move this from using the experimental repository to the unstable repository. We will move it up to production later. We don't have a full Debian 12 release of advligorts as we are not planning on supporting the rcg and real time code on Debian 12.
Dave and I will be watching the timings and checksums to make sure everything looks good. The IO setup is a bit different between h1daqfw2 and h1daqfw0 (fw2 writes to local disk, fw0 to a NAS).
As the new framewriter is a different piece of software, we track different metrics. So we will need an updated edc capture list and a daqd restart.
A note about what this system brings in.
* It automatically changes its configuration when there is a channel list. It is a step towards not having daqd restarts (those don't go away while we have the daqd in other places).
* When we restart the daqd we usually have a gap of several frames. We hide this from the outside world by having two frame writers and only restarting one at a time. This reconfigures the channel list and does not have gaps unless you actually restart the program.
* It is also being used in the ngdd work to write derived frame data, so that we have a common frame writer (we just change the input type).