exp real size, fix nopack measurement, time ranges

1 files changed, 32 insertions, 9 deletions

diff --git a/exp/Makefile b/exp/Makefile
index f49d2f1..b69b0bf 100644
--- a/exp/Makefile
+++ b/exp/Makefile
@@ -8,33 +8,38 @@ export LC_NUMERIC  :=
 export DNA_BACKUP  ?= ../dna-backup
 export DNA_PARAMS  ?= dna_params
 export MAX_VERSION ?= 5
-export COMMITS     ?= commits
+export COMMITS     ?= commits.daily
 DNA_4K             ?= dna_4k
 DNA_8K             ?= dna_8k
 NOPACK             ?= nopack
 export GIT_NOPACK  ?= $(abspath $(NOPACK))
+export REAL        ?= real
 export DIFFS       ?= diffs
 
 export REPO_PATH   ?= repo
 export GIT_PATH    ?= $(abspath git)
 GITC      := git -C $(REPO_PATH)
 
+# Number of initial commits to skip. This option has been added for the
+# Linux kernel repo, as the first commits are too far away date wise.
+SKIP_COMMITS ?= 30
+
 DNADIRS   := $(DNA_4K) $(DNA_8K)
-DATADIRS  := $(DNA_4K) $(DNA_8K) $(DIFFS) $(NOPACK)
+DATADIRS  := $(DNA_4K) $(DNA_8K) $(DIFFS) $(NOPACK) $(REAL)
 SIZEFILES := $(DATADIRS:%=%.size)
 
 .PHONY: all exp
 all exp: summary.csv $(SIZEFILES)
 	@echo "=============== SUMMARY ==============="
 	@head -n 1 $< | while IFS="," read -r $(DATADIRS); do \
-		printf "$(DATADIRS:%=\%13s)\n" $(DATADIRS:%=$$%); \
+		printf "$(DATADIRS:%=\%14s)\n" $(DATADIRS:%=$$%); \
 	done
 	@tail -n +2 $< | while IFS="," read -r $(DATADIRS); do \
-		printf "$(DATADIRS:%=\%'13d)\n" $(DATADIRS:%=$$%); \
+		printf "$(DATADIRS:%=\%'14d)\n" $(DATADIRS:%=$$%); \
 	done
 	@echo "================ TOTAL ================"
 	@for i in $(SIZEFILES); do cat $$i | paste -sd+ | bc; done | while read j; do \
-		printf "%'13d " $$j; \
+		printf "%'14d " $$j; \
 	done
 	@echo
 
@@ -42,12 +47,23 @@ summary.csv: $(SIZEFILES)
 	echo $(DATADIRS) | tr ' ' ',' > $@
 	paste -d "," $^ >> $@
 
-$(DNADIRS:%=%.size) $(NOPACK).size: %.size: %.versions run
+$(DNADIRS:%=%.size) $(REAL).size: %.size: %.versions run
 	rm -rf $@
 	for i in $</*; do \
 		cut -f1 $$i | paste -sd+ | bc >> $@; \
 	done
 
+$(NOPACK).size: $(NOPACK).versions
+	prev=/dev/null; for i in $</*; do \
+		diff -u0 $$prev $$i \
+		| sed -e '1,2d' -e '/^-/d' -e 's/^+//' -e '/^@@/d' \
+		| cut -f1 \
+		| paste -sd+ \
+		| bc \
+		>> $@; \
+		prev=$$i; \
+        done
+
 %.size: % run
 	find $< -type f -exec du -ba {} + | sort -k2 | cut -f1 > $@
 
@@ -65,7 +81,7 @@ $(DNADIRS:%=%.size) $(NOPACK).size: %.size: %.versions run
 	| sort -k2 \
 	> $@
 
-run: $(COMMITS) $(DNA_BACKUP) $(DNA_PARAMS) | $(DATADIRS) $(NOPACK).versions
+run: $(COMMITS) $(DNA_BACKUP) $(DNA_PARAMS) | $(DATADIRS) $(NOPACK).versions $(REAL).versions
 	rm -rf $(DATADIRS:%=%/*)
 	./exp.sh
 	touch $@
@@ -74,15 +90,22 @@ $(DNA_PARAMS):
 	echo "$(DNA_4K)	-c 4096" >> $@
 	echo "$(DNA_8K)	-c 8192" >> $@
 
-$(DATADIRS) $(NOPACK).versions:
+$(DATADIRS) $(NOPACK).versions $(REAL).versions:
 	mkdir $@
 
 $(DNA_BACKUP): .FORCE
 	@$(MAKE) -C $(@D) --no-print-directory $(@F)
 
-$(COMMITS): | repo git
+commits.monthly: commits.daily
+	sed -n '0~30p' $< > $@
+
+commits.weekly: commits.daily
+	sed -n '0~7p' $< > $@
+
+commits.daily: | repo git
 	$(GITC) log --reverse --no-merges --pretty=tformat:"%H	%as" \
 	| sort --unique --key=2 \
+	| tail -n +$(SKIP_COMMITS) \
 	> $@
 
 repo git &: