02:53 <mattip> _aegis_: the blake2b and blake2s implementations should be imported when you import hashlib.py in lib-python/3

02:54 <mattip> you should do that import once as part fo the build script since the first time they use cffi to compile

02:54 <mattip> post-translation

05:01 <_aegis_> oh I see modules were added to build_cffi_imports and the error was just confusing

06:15 <_aegis_> ok it's kinda confusing that build_cffi_imports throws the same errors

06:15 <_aegis_> but I see it works afterwards

09:36 <smvv> I made a small tool to visualize pypy trace logs: https://github.com/smvv/pypy-traceview/ It has some ideas from jitviewer (e.g. use objdump for disassembling). I couldn't get jitviewer to work on my machine, so I tried to build something simple. I've added some tests to the code base to prevent bitrot. What do you think?

10:06 <cfbolz> smvv: nice! I'll try it the next time I am reading some logs :-)

13:35 <Remi_M> smvv: looks awesome! I also sometimes struggled with getting jitviewer to work with recent or modified versions of pypy.

13:35 <Remi_M> It would be cool if you could link the traces and bridges, and maybe show a graphical representation of the trace-tree (similar to what vmprof can do (vmprof.com is down?)). I always got lost when I tried to navigate through the traces with jitviewer...

13:43 <arigato> yes, nice! my own feature request is to link the RPython IR with the actual assembler, using the numbers that appear as prefixes in both column

15:32 <smvv> @Remi_M do you have a screenshot or something that shows a trace-tree? I made a visualizer for spidermonkey's IR. The first screen was a graph where each node was a basic block. The nodes has a darker red when the node was more executed ("hot") than the others. Not sure if that is what you mean. I also thought about a simple filter (e.g. filenames, method names) that will only show the traces that have those names in it

15:36 <smvv> @Remi_M or do you mean this type of flamegraph? https://4.bp.blogspot.com/-n5LoH2hf7qI/WNvtNvIAbsI/AAAAAAAAAqc/zn0AXv8fkzIMQXWUwMLtLFpjochspz5MwCLcB/s1600/flamegraph.png

21:49 <mjacob> any news on the sprint front?

21:50 <mjacob> cfbolz: i remember you talked about a duesseldorf sprint

21:50 <cfbolz> mjacob: yes, I sent a pre-announcement to the mailing list with dates

21:50 <cfbolz> Should get to more info soon

21:51 <mjacob> ok, for some reason i missed that

21:51 <mjacob> did i miss any other sprint (pre-)announcement?

21:51 <cfbolz> Don't think so

21:52 <mjacob> cfbolz: so the date is fixed?

21:52 <cfbolz> Yes

21:52 <mjacob> ok, thanks for the information

21:54 <cfbolz> mjacob: are you planning to come?