Skip to contentSkip to conversation

pack-format: document missing fields

PATCH In review Alba Mendez via GitGitGadget 3 participants 3 messages Apr 9, 2020 – Apr 10, 2020 Documentation

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 pack-format: document missing fields

From: Alba Mendez <[email protected]>

Document missing fields in the description of the delta data, added at 011b648 (pack-format.txt: more details on pack file format, 2018-05-11).

Also, the description of object entries at the main section is a bit vague. There's an equivalent but more detailed description in the index section; since they are redundant, put that one in place of the first.

I have also expanded tabs to prevent alignment issues and rephrased a bit to make things clearer.

---
pack-format: document missing fields

Hello! These are some fixes & improvements for the pack-format documentation.

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-605%2Fmildsunrise%2Fpatch-1-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-605/mildsunrise/patch-1-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/605

 Documentation/technical/pack-format.txt | 75 ++++++++++---------------
 1 file changed, 31 insertions(+), 44 deletions(-)
Documentation/technical/pack-format.txt +31 -44
index d3a142c6520..b78e03137f4 100644
@@ -9,7 +9,7 @@ Git pack format
          The signature is: {'P', 'A', 'C', 'K'}
 
      4-byte version number (network byte order):
-	 Git currently accepts version number 2 or 3 but
+         Git currently accepts version number 2 or 3 but
          generates version 2 only.
 
      4-byte number of objects contained in the pack (network byte order)
@@ -20,19 +20,34 @@ Git pack format
    - The header is followed by number of object entries, each of
      which looks like this:
 
-     (undeltified representation)
-     n-byte type and length (3-bit type, (n-1)*7+4-bit length)
-     compressed data
-
-     (deltified representation)
-     n-byte type and length (3-bit type, (n-1)*7+4-bit length)
-     20-byte base object name if OBJ_REF_DELTA or a negative relative
-	 offset from the delta object's position in the pack if this
-	 is an OBJ_OFS_DELTA object
-     compressed delta data
+     packed object header:
+        1-byte size extension bit (MSB)
+            type (next 3-bit)
+            size0 (lower 4-bit)
+        n-byte sizeN (as long as MSB is set, each 7-bit)
+            size0..sizeN form 4+7+7+..+7 bit integer, size0
+            is the least significant part, and sizeN is the
+            most significant part.
+     packed object data:
+        If it is not DELTA, then deflated bytes (the
+            size above is the size before compression).
+        If it is REF_DELTA, then
+            20-byte base object name SHA-1.
+            deflated delta data (the size above is the
+                size of this data before compression).
+        If it is OFS_DELTA, then
+            n-byte offset (see below) interpreted as a
+                negative offset from the type-byte of
+                the header of the ofs-delta entry.
+            deflated delta data (the size above is the
+                size of this data before compression).
 
-     Observation: length of each object is encoded in a variable
-     length format and is not constrained to 32-bit or anything.
+     offset encoding:
+        n bytes with MSB set in all but the last one.
+        The offset is then the number constructed by
+        concatenating the lower 7 bit of each byte, and
+        for n >= 2 adding 2^7 + 2^14 + ... + 2^(7*(n-1))
+        to the result.
 
   - The trailer records 20-byte SHA-1 checksum of all of the above.
 
@@ -67,7 +82,9 @@ Ref-delta can also refer to an object outside the pack (i.e. the
 so-called "thin pack"). When stored on disk however, the pack should
 be self contained to avoid cyclic dependency.
 
-The delta data is a sequence of instructions to reconstruct an object
+The deflated delta data begins with two n-byte sizes: the size of
+the base object data, and the size of the reconstructed object data.
+What follows is a sequence of instructions to reconstruct the object
 from the base object. If the base object is deltified, it must be
 converted to canonical form first. Each instruction appends more and
 more data to the target object until it's complete. There are two
@@ -186,36 +203,6 @@ trailer	  | | packfile checksum              |
                   |
 Pack file entry: <+
 
-     packed object header:
-	1-byte size extension bit (MSB)
-	       type (next 3 bit)
-	       size0 (lower 4-bit)
-        n-byte sizeN (as long as MSB is set, each 7-bit)
-		size0..sizeN form 4+7+7+..+7 bit integer, size0
-		is the least significant part, and sizeN is the
-		most significant part.
-     packed object data:
-        If it is not DELTA, then deflated bytes (the size above
-		is the size before compression).
-	If it is REF_DELTA, then
-	  20-byte base object name SHA-1 (the size above is the
-		size of the delta data that follows).
-          delta data, deflated.
-	If it is OFS_DELTA, then
-	  n-byte offset (see below) interpreted as a negative
-		offset from the type-byte of the header of the
-		ofs-delta entry (the size above is the size of
-		the delta data that follows).
-	  delta data, deflated.
-
-     offset encoding:
-	  n bytes with MSB set in all but the last one.
-	  The offset is then the number constructed by
-	  concatenating the lower 7 bit of each byte, and
-	  for n >= 2 adding 2^7 + 2^14 + ... + 2^(7*(n-1))
-	  to the result.
-
-
 
 == Version 2 pack-*.idx files support packs larger than 4 GiB, and
    have some other reorganizations.  They have the format:

base-commit: 9fadedd637b312089337d73c3ed8447e9f0aa775

Signature · 1 line
gitgitgadget
#1 lore raw

Documentation/technical/pack-format.txt Git pack format · line 13

Quoted 4 lines · Alba Mendez via GitGitGadget

4-byte version number (network byte order):

Documentation/technical/pack-format.txt +1 -1
-	 Git currently accepts version number 2 or 3 but
+         Git currently accepts version number 2 or 3 but
          generates version 2 only.

Hmph, what is this hunk about?

Quoted 13 lines
Documentation/technical/pack-format.txt +1 -10
-     (undeltified representation)
-     n-byte type and length (3-bit type, (n-1)*7+4-bit length)
-     compressed data
-
-     (deltified representation)
-     n-byte type and length (3-bit type, (n-1)*7+4-bit length)
-     20-byte base object name if OBJ_REF_DELTA or a negative relative
-	 offset from the delta object's position in the pack if this
-	 is an OBJ_OFS_DELTA object
-     compressed delta data
 
 
+     packed object header:

Documentation/technical/pack-format.txt Git pack format

Quoted 10 lines
Documentation/technical/pack-format.txt +10 -0
+        1-byte size extension bit (MSB)
+            type (next 3-bit)
+            size0 (lower 4-bit)
+        n-byte sizeN (as long as MSB is set, each 7-bit)
+            size0..sizeN form 4+7+7+..+7 bit integer, size0
+            is the least significant part, and sizeN is the
+            most significant part.
+     packed object data:
+        If it is not DELTA, then deflated bytes (the
+            size above is the size before compression).

Correct.

Documentation/technical/pack-format.txt Git pack format · line 37

Quoted 4 lines
Documentation/technical/pack-format.txt +4 -0
+        If it is REF_DELTA, then
+            20-byte base object name SHA-1.
+            deflated delta data (the size above is the
+                size of this data before compression).

Correct.

Documentation/technical/pack-format.txt Git pack format · line 41

Quoted 4 lines
Documentation/technical/pack-format.txt +4 -0
+        If it is OFS_DELTA, then
+            n-byte offset (see below) interpreted as a
+                negative offset from the type-byte of
+                the header of the ofs-delta entry.

Correct, and "see below" here is very important.

Documentation/technical/pack-format.txt Git pack format · line 50

Quoted 9 lines
Documentation/technical/pack-format.txt +8 -0
+            deflated delta data (the size above is the
+                size of this data before compression).
 
+     offset encoding:
+        n bytes with MSB set in all but the last one.
+        The offset is then the number constructed by
+        concatenating the lower 7 bit of each byte, and
+        for n >= 2 adding 2^7 + 2^14 + ... + 2^(7*(n-1))
+        to the result.

Correct. Shouldn't we have a similar explanation for the object length varint above (in other words, how 4+7+7+..+7 bits are combined to form an integer)?

Documentation/technical/pack-format.txt Ref-delta can also refer to an object outside the pack (i.e. the · line 86

Quoted 7 lines
Documentation/technical/pack-format.txt +2 -1
@@ -67,7 +82,9 @@ Ref-delta can also refer to an object outside the pack (i.e. the
 so-called "thin pack"). When stored on disk however, the pack should
 be self contained to avoid cyclic dependency.
 
-The delta data is a sequence of instructions to reconstruct an object
+The deflated delta data begins with two n-byte sizes: the size of
+the base object data, and the size of the reconstructed object data.

Similarly, not just "n-byte size", but clarify that this is a pair of "varint" and specify which kind of "varint" it is (IIRC, this encoding did not use the "save-one-bit" trick the offset encoding uses).

Quoted 24 lines
Documentation/technical/pack-format.txt +0 -20
@@ -186,36 +203,6 @@ trailer	  | | packfile checksum              |
                   |
 Pack file entry: <+
 
-     packed object header:
-	1-byte size extension bit (MSB)
-	       type (next 3 bit)
-	       size0 (lower 4-bit)
-        n-byte sizeN (as long as MSB is set, each 7-bit)
-		size0..sizeN form 4+7+7+..+7 bit integer, size0
-		is the least significant part, and sizeN is the
-		most significant part.
-     packed object data:
-        If it is not DELTA, then deflated bytes (the size above
-		is the size before compression).
-	If it is REF_DELTA, then
-	  20-byte base object name SHA-1 (the size above is the
-		size of the delta data that follows).
-          delta data, deflated.
-	If it is OFS_DELTA, then
-	  n-byte offset (see below) interpreted as a negative
-		offset from the type-byte of the header of the
-		ofs-delta entry (the size above is the size of
-		the delta data that follows).

Documentation/technical/pack-format.txt trailer | | packfile checksum | · line 218

Quoted 10 lines
Documentation/technical/pack-format.txt +0 -10
-	  delta data, deflated.
-
-     offset encoding:
-	  n bytes with MSB set in all but the last one.
-	  The offset is then the number constructed by
-	  concatenating the lower 7 bit of each byte, and
-	  for n >= 2 adding 2^7 + 2^14 + ... + 2^(7*(n-1))
-	  to the result.
-
-

Hmph, I kept saying "Correct" above, but it seems that the correctness come from the original. So, why is this patch so big?

As far as I can tell, the only thing it improved (content-wise) was to say that the delta data has two varints that lets reader to validate the length of the original contents that the delta applies to, and to allocate the buffer to hold the result before applying the delta. Puzzled.

Thanks.

#2 lore raw
Quoted 8 lines · Junio C Hamano

Documentation/technical/pack-format.txt Git pack format · line 13

Quoted 4 lines · Alba Mendez via GitGitGadget

4-byte version number (network byte order):

Documentation/technical/pack-format.txt +1 -1
-	 Git currently accepts version number 2 or 3 but
+         Git currently accepts version number 2 or 3 but
          generates version 2 only.

Hmph, what is this hunk about?

The original line is "<TAB><SPACE>Git", she changed to 9 <SPACE> to match with other lines.
I guess she set tabstop=4, this paragraph will be mis-aligned on that.

Quoted 9 lines

Documentation/technical/pack-format.txt Git pack format · line 31

Quoted 9 lines
Documentation/technical/pack-format.txt +0 -9
-     (undeltified representation)
-     n-byte type and length (3-bit type, (n-1)*7+4-bit length)
-     compressed data
-
-     (deltified representation)
-     n-byte type and length (3-bit type, (n-1)*7+4-bit length)
-     20-byte base object name if OBJ_REF_DELTA or a negative relative
-	 offset from the delta object's position in the pack if this
-	 is an OBJ_OFS_DELTA object

Alba, if you gonna change the above <TAB> to 8 <SPACE> above, you may want to change here, too.
This paragraph was written with tabstop=4 in mind.
And it's in the context of below hunk.

I'm not sure if changing <TAB> to 8 <SPACE> was expected in this file because <TAB> and <SPACE> is also mixed in "Pack file entry" section.

Signature · 1 line
Danh