Spanner: Read single key with index

[chemelnucfin]

I'm trying for smaller PRs to simplify debugging/code writing for other issues.

[dhermes]

I'll let @tseaver weigh in, but it seems this is lacking an explanation of

• Why the timeout had to go up from 90 to 120
• Why CREATE INDEX name ON contacts(first_name, last_name); was added to the DDL
• Why ROW_COUNT was lowered from 4000 to 3000

[chemelnucfin]

Please see #4386 for the CREATE INDEX reasoning, and ROW_COUNT needs to be lowered because apparently 20000 mutations is the cap, and adding an additional INDEX apparently gets to the cap. Short story is that instead of building a temp and removing it, it was added to the fixture for easier manipulation. The timeout at 90 would fail occasionally for me running the tests. 120 so far has never failed for me.

…

On Thu, Nov 16, 2017 at 1:52 PM, Danny Hermes ***@***.***> wrote: I'll let @tseaver <https://github.com/tseaver> weigh in, but it seems this is lacking an explanation of - Why the timeout had to go up from 90 to 120 - Why CREATE INDEX name ON contacts(first_name, last_name); was added to the DDL - Why ROW_COUNT was lowered from 4000 to 3000 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#4403 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADzDDDl9zzohA_2PFXV4U4OUpM_rE4xjks5s3K6hgaJpZM4QhIcm> .