exp: repo 4k, better summary format, fix diff sizes

2 files changed, 61 insertions, 25 deletions

diff --git a/exp/Makefile b/exp/Makefile
index aaf377c..8fb4ce0 100644
--- a/exp/Makefile
+++ b/exp/Makefile
@@ -1,48 +1,62 @@
 REPO_URL  := https://club1.fr/~nicolas/git/dna-backup/
 
+SHELL := /bin/bash
+
+# Reset LC_NUMERIC to make printf thousands separator work as expected
+export LC_NUMERIC  :=
+
 export DNA_BACKUP  ?= ../dna-backup
 export MAX_VERSION ?= 5
 export COMMITS     ?= commits
-export BACKUP      ?= backup
+export DNA_4K      ?= dna_4k
+export DNA_8K      ?= dna_8k
 export DIFFS       ?= diffs
 
 export REPO_PATH   ?= repo
 GIT_PATH  := git
 GITC      := git -C $(REPO_PATH)
 
-DATADIRS  := $(BACKUP) $(DIFFS)
-SIZEFILES := $(DATADIRS:%=%-size)
+DNADIRS   := $(DNA_4K) $(DNA_8K)
+DATADIRS  := $(DNA_4K) $(DNA_8K) $(DIFFS)
+SIZEFILES := $(DATADIRS:%=%.size)
 
 .PHONY: all exp
 all exp: summary.csv $(SIZEFILES)
-	@echo "============== SUMMARY =============="
-	@cat $< | tr ',' '\t'
-	@echo "==============  TOTAL  =============="
-	@for i in $(SIZEFILES); do cat $$i | paste -sd+ | bc; done | tr '\n' '\t'
+	@echo "=============== SUMMARY ==============="
+	@head -n 1 $< | while IFS="," read -r $(DATADIRS); do \
+		printf "$(DATADIRS:%=\%13s)\n" $(DATADIRS:%=$$%); \
+	done
+	@tail -n +2 $< | while IFS="," read -r $(DATADIRS); do \
+		printf "$(DATADIRS:%=\%'13d)\n" $(DATADIRS:%=$$%); \
+	done
+	@echo "================ TOTAL ================"
+	@for i in $(SIZEFILES); do cat $$i | paste -sd+ | bc; done | while read j; do \
+		printf "%'13d " $$j; \
+	done
 	@echo
 
 summary.csv: $(SIZEFILES)
 	echo $(DATADIRS) | tr ' ' ',' > $@
 	paste -d "," $^ >> $@
 
-backup-size: versions
+$(DNADIRS:%=%.size): %.size: %.versions
 	rm -rf $@
 	for i in $</*; do \
 		cut -f1 $$i | paste -sd+ | bc >> $@; \
 	done
 
-diffs-size: diffs run
-	find $< -type f -exec du -ba {} + | cut -f1 > $@
+%.size: % run
+	find $< -type f -exec du -ba {} + | sort -k2 | cut -f1 > $@
 
-versions: results
+%.versions: %.results
 	rm -rf $@
 	mkdir -p $@
-	for i in backup/*; do \
+	for i in $*/*; do \
 		v=`echo $$i | cut -d / -f2`; \
 		grep $$i $< > $@/$$v ; \
 	done
 
-results: backup run
+%.results: % run
 	find $< -type f -exec du -ba {} + \
 	| grep -v hashes \
 	| sort -k2 \
@@ -53,7 +67,7 @@ run: $(COMMITS) $(DNA_BACKUP) | $(DATADIRS)
 	./exp.sh
 	touch $@
 
-backup diffs:
+$(DATADIRS):
 	mkdir $@
 
 $(DNA_BACKUP): .FORCE
@@ -75,9 +89,9 @@ clean: mostlyclean
 	rm -f $(COMMITS)
 
 mostlyclean: resultsclean
-	rm -rf $(DATADIRS) versions run
+	rm -rf $(DATADIRS) $(DNADIRS:%=%.versions) run
 
 resultsclean:
-	rm -f results summary.csv $(SIZEFILES)
+	rm -f summary.csv $(SIZEFILES)
 
 .FORCE: ;
diff --git a/exp/exp.sh b/exp/exp.sh
index f7a603c..8a7af40 100755
--- a/exp/exp.sh
+++ b/exp/exp.sh
@@ -5,7 +5,8 @@
 # - REPO_PATH: the path of the repo the experiment is based on
 # - MAX_VERSION: the max number for versions for the experiment
 # - COMMITS: the name of the file that contains the lists of versions
-# - BACKUP: the path fo the dna-backup dir
+# - DNA_4K: the path fo the dna-backup dir with 4K chunksize
+# - DNA_8K: the path fo the dna-backup dir with 8K chunksize
 # - DIFFS: the path of the git diff dir
 
 log() {
@@ -22,27 +23,38 @@ last=$(tail --lines=1 $COMMITS | cut -f1)
 i=0
 cat $COMMITS | while read line
 do
+	# Get hash
 	hash=$(echo "$line" | cut -f1)
 
+	# Check out repo
 	log "check out $hash"
 	$GITC checkout $hash 2> $OUT \
 	|| (log "error checking out"; cat $OUT; exit 1)
 
-	log "create diff for this version"
+	# Create git diff for this version
+	log "create git diff for this version"
+	diff=$(printf "%s/%05d.diff.gz" $DIFFS $i)
 	$GITC diff --minimal --binary --unified=0 -l0 $prev \
 	| gzip \
-	> "$DIFFS/$i.diff.gz"
+	> $diff
 
-	log "create backup for this version"
-	$DNA_BACKUP commit -v 2 $REPO_PATH $BACKUP
+	# Create 4k dna backup for this version
+	log "create 4k dna backup for this version"
+	$DNA_BACKUP commit -v 2 -c 4096 $REPO_PATH $DNA_4K
+
+	# Create 8k dna backup for this version
+	log "create 8k dna backup for this version"
+	$DNA_BACKUP commit -v 2 $REPO_PATH $DNA_8K
 
 	if [[ $(( $i % 4 )) == 0 ]]
 	then
-		log "restore from diffs"
+		# Check restore from git diffs
+		log "restore from git diffs"
 		TEMP=$(mktemp -d)
 		for n in $(seq 0 $i)
 		do
-			cat "$DIFFS/$n.diff.gz" \
+			diff=$(printf "%s/%05d.diff.gz" $DIFFS $n)
+			cat $diff \
 			| gzip --decompress \
 			| git -C $TEMP apply --binary --unidiff-zero --whitespace=nowarn -
 		done
@@ -52,9 +64,19 @@ do
 		|| log "git patchs restore do not match source"
 		rm -rf $TEMP
 
-		log "restore from backup"
+		# Check restore from 4k dna backup
+		log "restore from 4k dna backup"
+		TEMP=$(mktemp -d)
+		$DNA_BACKUP restore -v 2 -c 4096 $DNA_4K $TEMP
+		log "check restore from backup"
+		diff --brief --recursive $REPO_PATH $TEMP \
+		|| log "dna backup restore do not match source"
+		rm -rf $TEMP
+
+		# Check restore from 8k dna backup
+		log "restore from 8k dna backup"
 		TEMP=$(mktemp -d)
-		$DNA_BACKUP restore -v 2 $BACKUP $TEMP
+		$DNA_BACKUP restore -v 2 $DNA_8K $TEMP
 		log "check restore from backup"
 		diff --brief --recursive $REPO_PATH $TEMP \
 		|| log "dna backup restore do not match source"