jerch

@dankamongmen You prolly got mail :smile_cat:

> I also need new oil for my poor satellites first. But a few things for now: > > * looking at this thread clearly shows that just that one...

Eww, yes can repro that with `nano`. To me this seems rather awkward, maybe it is a workaround for some older TE bugs? Following default buffer switch behavior I would...

> As I mentioned above, I am going to open a separate discussion about the cache management policy of the draft proposal. Sure, just one thing - I am not...

@christianparpart Sounds good to me. :+1: Imho the placeholder thingy is helpful for peeps to get at least a visual feedback, why there is some room in the scrollback, that...

@wez Good point, thanks for bringing this up. @Tyriar has worked in that field for xterm.js and even proposed an OSC sequence scheme for accessibility annotations to the terminal-wg. This...

Yepp, anything beyond the base64 alphabet chars.

Yeah, you prolly can simply aggregate error conditions in another v128 variable, which only needs to be eval'ed once at the end. Would be enough for a good vs. bad...

Hmm I get almost no speed benefit with SIMD + error check compared to a uint32 LUT variant (its ~1.5 GB/s SIMD vs ~1.4 GB/s LUT). Still investigating whether SIMD...

Update: With Mulas [pshufb variant](http://0x80.pl/notesen/2016-01-17-sse-base64-decoding.html#vector-lookup-pshufb) I get it to >2.5 GB/s (correct error detection still untested). All other variants are much worse. The higher bit variants by Lemire are not...