Tom Bishop

@macchiati @srl295 this is the same as https://github.com/unicode-org/cldr/pull/1911 except that some lines tend to be longer

"Error: /home/runner/work/cldr/cldr/tools/cldr-apps/src/main/java/org/unicode/cldr/util/CLDRConfigImpl.java:[19,26] package javax.servlet.http does not exist" ?

@srl295 I've been experimenting with your idea locally. So far it seems that sleeping for ten seconds has no effect on garbage collection. With multiple threads running, gc may not...

"the heap is full of reachable objects that won't get gc'ed" -- maybe you're right "until other processing finishes and releases memory" -- during Summary, it seems like the only...

@srl295 i'm seeing mixed results. 7 GB does seem to have helped the situation on cldr-smoke, since i was able to make a snapshot there manually as admin... however, even...

@srl295 "Making a snapshot should be uncontested though, right? VettingViewer's computeAll() is called from a single thread." -- i think that's right... so with the current code, the compute() method...

I've been experimenting with a new org.unicode.cldr.util.MemoryHelper for the reportFreeMemory method if the work for Summary is all being done in a single thread running computeAll() then it seems that...

FWIW, to me the most important problems a formatter should fix are: (1) irregular indentation (spaces/tabs at starts of lines) (2) more or less than one space where one space...

> If people think all we need to fix is indentation and internal spaces, I'm skeptical it's worth doing anything at all. > Some of the CLDR code is hard...

I did add some comments in https://github.com/unicode-org/cldr/pull/880 If the reformatter has to divide and join lines, it would be better to err on the side of dividing. Joining is particularly...