| cesar | octavius: Indeed, GTKWave will silently drop any trace not present in the VCD file. SInce my API just writes the gtkw file, withput looking at the VCD, it can't give a warning either. | 10:12 |
|---|---|---|
| cesar | GTKWave could at least give a warning on the console. | 10:13 |
| cesar | Now, once the trace is dropped by GTKWave, it seems its color is given to the next trace. I hadn't noticed this before. I think you really found a bug of GTKWave! | 10:16 |
| cesar | ... or at least an instance of undefined behavior. | 10:17 |
| cesar | By the way, thanks for fixing the tutorial page. | 10:18 |
| cesar | Also, feel free to fix p_shift_i to p_i_shift. The automated search and replace, on the latest signal renaming event, likely missed this signal. | 10:24 |
| octavius | Thank you cesar for making this API wrapper! I was initially confused and a little frustrated, but by actually following the tutorial (*who would've thought...*) I was able to implement gtkw doc generation in my jtag test code | 11:03 |
| octavius | I'll update that shift signal name too | 11:03 |
| lkcl | cool! | 11:19 |
| lkcl | now all it will take is for me to actually write one too (doh) | 11:31 |
| lkcl | hmmm, *thinks*... | 11:32 |
| lkcl | how about we add this as a feature in nmigen? and associated tree/node-walker which collates gtkw information? | 11:32 |
| octavius | "and associated tree/node-walker which collates gtkw information?" I understood was words, I think XD | 12:29 |
| octavius | I'd like to make the "traces" list be generated automatically based on the nmigen signal names, and I had some ideas for how it could be done. Having a tree containing all the relevant info would probably make the job easier though | 12:31 |
| octavius | But I keep getting distracted, I'll focus on the pinmux for now | 12:31 |
| lkcl | basically, the idea is, each Module has a get_gtkw_css_info() function and if it is not present we don't care | 13:55 |
| lkcl | then a tree/node-walker of the entire set of Modules goes "oh, do you have a get_gtkw_css_info() function? if so let me just call that for you and add whatever-you-return to the gtwk-css 'thing' i am currently creating for you" | 14:00 |
| lkcl | otherwise you have to drill down manually into the structure of your HDL modules, hunting for things that, in your *next* design are in completely different places | 14:03 |
| cesar | There is a discussion in Amaranth for adding generic metadata to signals, so one could mark a signal as "please export me to gtkw". See: https://github.com/amaranth-lang/amaranth/pull/624 | 14:20 |
| lkcl | ahh i was looking for that. | 14:52 |
| lkcl | an advancement on it is to auto-propagate those down a netlist | 14:52 |
| lkcl | so that you can track an entire netlist rather than just a Signal | 14:53 |
| octavius | This is really cool actually | 19:00 |
| lkcl | oleeee i have a *nmigen* peripheral fabric for microwatt and libre-soc, verilator simulation | 20:09 |
| lkcl | it's the hello_world.bin | 20:09 |
| programmerjake | lkcl meeting | 22:06 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!