Michael Stapelberg

Our hand-written marshal code is about 10x faster (89 ns/op) than `proto.Marshal` (1102 ns/op), so we’ll stick with it.

The code I produced up until now can be found in https://github.com/robustirc/robustirc/tree/outputstream-simplify, but I’ll take a break from this. See the commit message for the currently known issues with that...

One more thing to note: the usage pattern actually is the same for all of our stores now (ircstore, raftstore, outputstream), I think. Possibly we should use the same implementation...

Thanks for the fix! Unfortunately, the algorithm still doesn’t seem entirely correct (tested with v0.0.2-0.20200403031649-7456955b3f48). Here are a few more samples for input where it spits out the wrong result:...

However: ``` % perl -MText::Levenshtein::Damerau -E 'say Text::Levenshtein::Damerau::edistance("3AWPLE", "9C9TEL")' 5 ``` Are you saying the Perl implementation is also incorrect?

> Here is the list of what we consider "system" libraries that we simply assume to be "there" on every target system: https://github.com/AppImage/pkg2appimage/blob/master/excludelist Is there a test program I can...

No, there seems to be a quoting issue: it looks like the list of libraries is interpreted as a single library: ``` xps ~/src % LIBS=$(cat excludelist | grep -v...

I’m seeing the same thing. Chrome works fine, in Firefox scrolling is reversed. This also happens on a Mac, not just on Linux.

It looks like the edittable plugin is using an old version of https://github.com/handsontable/handsontable. Perhaps updating to a more recent version includes a fix, or at least allows us to report...

Oops! Yeah, reducing CPU usage would be great. Improvements welcome :)