wow, "make test" is slow...

[Eric Wong]

It's been some time, but it took me ~428s for "make test" with
SQLite. I know it was slow before, but over 7 minutes?
I suppose some of the CPU vulnerabilities over the years have
slowed things down.

No wonder I've tended to work on other things... I've been
spoiled by 10-20s test suite times in my other Perl project,
fully utilizing the 4 cores I have...

Haven't dug into it (nor do I really want to), but I guess lack
of parallelization could be some of the problem.

I'm using the following Makefile to wrap tests for SQLite (I
always run tests with eatmydata (LD_PRELOAD to disable
fsync/sync/O_SYNC).
---8<---
# eatmydata make -f .git/test.mk sqlite

PATH := ../MogileFS-Utils/:$(PATH)
PERL5LIB := ../perl-MogileFS-Client/lib
export PATH PERL5LIB

sqlite:
$(MAKE) test MOGTEST_DBUSER=mogile \
MOGTEST_DBNAME=tmp_mogiletest.sqlite3 \
MOGTEST_DBTYPE=SQLite \
MOGTEST_DB_PASS=

# there's similar targets for postgres and mysql, but
# I haven't had servers installed/running for those in
# many, many years.