From 2df30c27ead445724d066fc7df301be4d22f3c44 Mon Sep 17 00:00:00 2001
From: n-peugnet <n.peugnet@free.fr>
Date: Fri, 8 Oct 2021 19:45:21 +0200
Subject: add git nopack to exp

---
 README.md               |  2 +-
 docs/note-2021-10-08.md | 20 ++++++++++++++++++++
 exp/Makefile            | 15 +++++++++------
 exp/exp.sh              | 29 +++++++++++++++++++++++++++++
 4 files changed, 59 insertions(+), 7 deletions(-)
 create mode 100644 docs/note-2021-10-08.md

diff --git a/README.md b/README.md
index e43c327..30774c9 100644
--- a/README.md
+++ b/README.md
@@ -14,7 +14,7 @@ sur un support de stockage classique : le _repo_.
  | +---------+          +---------+ |          +-----------+
  | |         |          |         | |          |           |
  | | Source  |--------->|  Repo   |----------->| DNA-Drive |
- | |         |  COMMIT  |         | | SYMTHESE |           |
+ | |         |  COMMIT  |         | | SYNTHESE |           |
  | +---------+          +---------+ |          +-----------+
  |                                  |
  | Ordinateur                       |
diff --git a/docs/note-2021-10-08.md b/docs/note-2021-10-08.md
new file mode 100644
index 0000000..fc26758
--- /dev/null
+++ b/docs/note-2021-10-08.md
@@ -0,0 +1,20 @@
+Comparison with git basic
+=========================
+
+```
+=============== SUMMARY ===============
+       dna_4k        dna_8k         diffs        nopack
+   85,062,292    77,288,903    59,771,518    83,421,168
+  113,976,574   108,519,248    75,958,096   126,395,519
+   40,450,844    43,371,572    87,905,291    62,515,404
+   13,699,644    17,821,696     7,398,260    47,162,686
+   39,029,362    40,321,219    28,806,094    55,473,207
+    9,335,530    11,047,234    20,915,481    32,021,383
+   16,891,778    20,219,859    12,926,085    42,470,711
+    9,764,957    11,381,646    13,592,417    28,181,096
+        2,597       131,258         1,395        55,764
+       93,015       243,287       104,610     2,739,143
+        9,331         7,919         2,607       102,283
+================ TOTAL ================
+  328,315,924   330,353,841   307,381,854   480,538,364 
+```
diff --git a/exp/Makefile b/exp/Makefile
index 8fb4ce0..4df2bb6 100644
--- a/exp/Makefile
+++ b/exp/Makefile
@@ -11,13 +11,15 @@ export COMMITS     ?= commits
 export DNA_4K      ?= dna_4k
 export DNA_8K      ?= dna_8k
 export DIFFS       ?= diffs
+NOPACK             ?= nopack
+export GIT_NOPACK  ?= $(abspath $(NOPACK))
 
 export REPO_PATH   ?= repo
-GIT_PATH  := git
+export GIT_PATH    ?= $(abspath git)
 GITC      := git -C $(REPO_PATH)
 
 DNADIRS   := $(DNA_4K) $(DNA_8K)
-DATADIRS  := $(DNA_4K) $(DNA_8K) $(DIFFS)
+DATADIRS  := $(DNA_4K) $(DNA_8K) $(DIFFS) $(NOPACK)
 SIZEFILES := $(DATADIRS:%=%.size)
 
 .PHONY: all exp
@@ -39,7 +41,7 @@ summary.csv: $(SIZEFILES)
 	echo $(DATADIRS) | tr ' ' ',' > $@
 	paste -d "," $^ >> $@
 
-$(DNADIRS:%=%.size): %.size: %.versions
+$(DNADIRS:%=%.size) $(NOPACK).size: %.size: %.versions run
 	rm -rf $@
 	for i in $</*; do \
 		cut -f1 $$i | paste -sd+ | bc >> $@; \
@@ -62,12 +64,12 @@ $(DNADIRS:%=%.size): %.size: %.versions
 	| sort -k2 \
 	> $@
 
-run: $(COMMITS) $(DNA_BACKUP) | $(DATADIRS)
+run: $(COMMITS) $(DNA_BACKUP) | $(DATADIRS) $(NOPACK).versions
 	rm -rf $(DATADIRS:%=%/*)
 	./exp.sh
 	touch $@
 
-$(DATADIRS):
+$(DATADIRS) $(NOPACK).versions:
 	mkdir $@
 
 $(DNA_BACKUP): .FORCE
@@ -89,7 +91,8 @@ clean: mostlyclean
 	rm -f $(COMMITS)
 
 mostlyclean: resultsclean
-	rm -rf $(DATADIRS) $(DNADIRS:%=%.versions) run
+	rm -rf $(DATADIRS) $(DATADIRS:%=%.versions)
+	rm -f run $(DATADIRS:%=%.results)
 
 resultsclean:
 	rm -f summary.csv $(SIZEFILES)
diff --git a/exp/exp.sh b/exp/exp.sh
index 8a7af40..d5cc4dd 100755
--- a/exp/exp.sh
+++ b/exp/exp.sh
@@ -3,19 +3,32 @@
 # This script expects the following variables to be exported:
 # - DNA_BACKUP: the path to dna-backup binary
 # - REPO_PATH: the path of the repo the experiment is based on
+# - GIT_PATH: the path of the repo git-dir
 # - MAX_VERSION: the max number for versions for the experiment
 # - COMMITS: the name of the file that contains the lists of versions
 # - 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
+# - GIT_NOPACK: the path of the git nopack dir
 
 log() {
 	echo -e "\033[90m$(date +%T.%3N)\033[0m" $*
 }
 
+set-git-dir() {
+	echo gitdir: $1 > $REPO_PATH/.git
+}
+
 GITC="git -C $REPO_PATH"
 OUT=/tmp/dna-backup-exp-out
 
+# Init git nopack dir
+rm $REPO_PATH/.git
+$GITC init --separate-git-dir=$GIT_NOPACK
+$GITC --git-dir=$GIT_NOPACK config gc.auto 0
+set-git-dir $GIT_PATH
+nopack_prev=0
+
 # "empty tree" commit
 prev="4b825dc642cb6eb9a060e54bf8d69288fbee4904"
 last=$(tail --lines=1 $COMMITS | cut -f1)
@@ -38,6 +51,22 @@ do
 	| gzip \
 	> $diff
 
+	# Create git nopack for this version
+	log "create git nopack for this version"
+	set-git-dir $GIT_NOPACK
+	$GITC add .
+	$GITC commit -m $hash &> $OUT \
+	|| (log "error commiting to nopack"; cat $OUT; exit 1)
+	ls $GIT_NOPACK/objects/pack
+	find $GIT_NOPACK -type f -exec du -ba {} + \
+	| grep -v /logs/ \
+	| cut -f1 \
+	| paste -sd+ \
+	| xargs -i echo {} - $nopack_prev \
+	| bc \
+	> $(printf "%s.versions/%05d" $GIT_NOPACK $i)
+	set-git-dir $GIT_PATH
+
 	# 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
-- 
cgit v1.2.3