delta: exclude base candidate much smaller than the target
If a revision's full text is that much bigger than a base candidate full text,
we no longer consider that candidate.
This solves a pathological case we encountered on a very specify repository.
It contains a long series of changesets with a very small manifest (one file)
co-existing with others changesets using a very large manifest.
Without this filtering, we ended up considering a large number of tiny full
snapshots as a potential base. It resulted in very large delta (the size of
the full text) and mercurial spending 99% of its time compressing these
deltas.
The timing of a commit moved from about 400s to about 10s (still slow, but not
ridiculously slow).
SOURCES=$(notdir $(wildcard ../mercurial/help/*.[0-9].txt))
MAN=$(SOURCES:%.txt=%)
HTML=$(SOURCES:%.txt=%.html)
GENDOC=gendoc.py ../mercurial/commands.py ../mercurial/help.py \
../mercurial/help/*.txt ../hgext/*.py ../hgext/*/__init__.py
PREFIX=/usr/local
MANDIR=$(PREFIX)/share/man
INSTALL=install -c -m 644
PYTHON=python
RSTARGS=
export HGENCODING=UTF-8
all: man html
man: $(MAN)
html: $(HTML)
common.txt $(SOURCES) $(SOURCES:%.txt=%.gendoc.txt): $(GENDOC)
${PYTHON} gendoc.py "$(basename $@)" > $@.tmp
mv $@.tmp $@
%: %.txt %.gendoc.txt common.txt
$(PYTHON) runrst hgmanpage $(RSTARGS) --halt warning \
--strip-elements-with-class htmlonly $*.txt $*
%.html: %.txt %.gendoc.txt common.txt
$(PYTHON) runrst html $(RSTARGS) --halt warning \
--link-stylesheet --stylesheet-path style.css $*.txt $*.html
MANIFEST: man html
# tracked files are already in the main MANIFEST
$(RM) $@
for i in $(MAN) $(HTML); do \
echo "doc/$$i" >> $@ ; \
done
install: man
for i in $(MAN) ; do \
subdir=`echo $$i | sed -n 's/^.*\.\([0-9]\)$$/man\1/p'` ; \
mkdir -p "$(DESTDIR)$(MANDIR)"/$$subdir ; \
$(INSTALL) $$i "$(DESTDIR)$(MANDIR)"/$$subdir ; \
done
clean:
$(RM) $(MAN) $(HTML) common.txt $(SOURCES) $(SOURCES:%.txt=%.gendoc.txt) MANIFEST