Skip to contentSkip to conversation

support non-negative float number in git-repack --max-pack-size

suggestion In review Bagas Sanjaya 4 participants 9 messages Jun 7, 2021 – Jun 18, 2021 Documentation

lore
Share & tools
lore
#0 lore raw

Hi,

I would like to create packfiles with charm-numbered size (that is for example use 49.99M instead of 50M) with git-repack:

$ git repack --max-pack-size=49.99M -a -d

But Git didn't support it:

Quoted 1 line

error: option `max-pack-size' expects a non-negative integer value with an optional k/m/g suffix

The workaround was scaling down to kibibytes:

$ git repack --max-pack-size=52418K -a -d

But the workaround is a rather convoluted to me, because I must convert mebibytes (MiB) to kibibytes (KiB). I had to multiply the desired packfile size by 1024, as opposed to by 1000 in familiar size notation (kilobytes [KB] and megabytes [MB]).

It would be nice if non-negative floating-point number can be allowed in --max-pack-size option, so that many users don't have to scale down size notation like above.

PS: charm numbers are most often used in pricing, because it's almost used everywhere (part of psychological pricing).

Thanks.

Signature · 1 line
An old man doll... just what I always wanted! - Clara
#1 lore raw
Quoted 4 lines · Bagas Sanjaya

I would like to create packfiles with charm-numbered size (that is for example use 49.99M instead of 50M) with git-repack:

$ git repack --max-pack-size=49.99M -a -d

The parser for numbers with units is shared by many options and config variables. In general, I'm not really opposed to allowing floating point values which get rounded to the nearest byte for any of them. So no objection for the general feature if somebody wants to implement it.

But I would note that --max-pack-size is almost never a good idea in general. I don't know what you think it's accomplishing, but it is probably making your repository larger than it needs to be, as well as less performant.

-Peff

#2 lore raw
Quoted 12 lines · Jeff King
Quoted 4 lines · Bagas Sanjaya

I would like to create packfiles with charm-numbered size (that is for example use 49.99M instead of 50M) with git-repack:

$ git repack --max-pack-size=49.99M -a -d

The parser for numbers with units is shared by many options and config variables. In general, I'm not really opposed to allowing floating point values which get rounded to the nearest byte for any of them. So no objection for the general feature if somebody wants to implement it.

Quoted 4 lines

But I would note that --max-pack-size is almost never a good idea in general. I don't know what you think it's accomplishing, but it is probably making your repository larger than it needs to be, as well as less performant.

Perhaps a doc update is in order? It might have been cute and superficially useful to be able to cut packfiles in 650M chunks to fit on a CD-ROM, but I suspect that it would be more useful to feed a single large file to a generic multi-volume archive tool and let it split it to fit the physical volume the tool deals with.

#3 lore raw

PATCH 1/1 doc: warn people against --max-pack-size

Quoted 19 lines · Junio C Hamano
Quoted 12 lines · Jeff King
Quoted 4 lines · Bagas Sanjaya

I would like to create packfiles with charm-numbered size (that is for example use 49.99M instead of 50M) with git-repack:

$ git repack --max-pack-size=49.99M -a -d

The parser for numbers with units is shared by many options and config variables. In general, I'm not really opposed to allowing floating point values which get rounded to the nearest byte for any of them. So no objection for the general feature if somebody wants to implement it.

Quoted 4 lines

But I would note that --max-pack-size is almost never a good idea in general. I don't know what you think it's accomplishing, but it is probably making your repository larger than it needs to be, as well as less performant.

Quoted 5 lines

Perhaps a doc update is in order? It might have been cute and superficially useful to be able to cut packfiles in 650M chunks to fit on a CD-ROM, but I suspect that it would be more useful to feed a single large file to a generic multi-volume archive tool and let it split it to fit the physical volume the tool deals with.

Yeah. Let's do that while we're thinking about it. Here's what I came up with.

-- >8 --
Subject: [PATCH] doc: warn people against --max-pack-size

This option is almost never a good idea, as the resulting repository is larger and slower (see the new explanations in the docs).

I outlined the potential problems. We could go further and make the option harder to find (or at least, make the command-line option descriptions a much more terse "you probably don't want this; see pack.packsizeLimit for details"). But this seems like a minimal change that may prevent people from thinking it's more useful than it is.

---
 Documentation/config/pack.txt      | 23 +++++++++++++++++------
 Documentation/git-pack-objects.txt |  6 +++---
 Documentation/git-repack.txt       |  4 +++-
 3 files changed, 23 insertions(+), 10 deletions(-)
Documentation/config/pack.txt +17 -6
index c0844d8d8e..763f7af7c4 100644
@@ -99,12 +99,23 @@ pack.packSizeLimit::
 	packing to a file when repacking, i.e. the git:// protocol
 	is unaffected.  It can be overridden by the `--max-pack-size`
 	option of linkgit:git-repack[1].  Reaching this limit results
-	in the creation of multiple packfiles; which in turn prevents
-	bitmaps from being created.
-	The minimum size allowed is limited to 1 MiB.
-	The default is unlimited.
-	Common unit suffixes of 'k', 'm', or 'g' are
-	supported.
+	in the creation of multiple packfiles.
++
+Note that this option is rarely useful, and may result in a larger total
+on-disk size (because Git will not store deltas between packs), as well
+as worse runtime performance (object lookup within multiple packs is
+slower than a single pack, and optimizations like reachability bitmaps
+cannot cope with multiple packs).
++
+If you need to actively run Git using smaller packfiles (e.g., because your
+filesystem does not support large files), this option may help. But if
+your goal is to transmit a packfile over a medium that supports limited
+sizes (e.g., removable media that cannot store the whole repository),
+you are likely better off creating a single large packfile and splitting
+it using a generic multi-volume archive tool (e.g., Unix `split`).
++
+The minimum size allowed is limited to 1 MiB. The default is unlimited.
+Common unit suffixes of 'k', 'm', or 'g' are supported.
 
 pack.useBitmaps::
 	When true, git will use pack bitmaps (if available) when packing
Documentation/git-pack-objects.txt +3 -3
index 25d9fbe37a..dbfd1f9017 100644
@@ -128,10 +128,10 @@ depth is 4095.
 	into multiple independent packfiles, each not larger than the
 	given size. The size can be suffixed with
 	"k", "m", or "g". The minimum size allowed is limited to 1 MiB.
-	This option
-	prevents the creation of a bitmap index.
 	The default is unlimited, unless the config variable
-	`pack.packSizeLimit` is set.
+	`pack.packSizeLimit` is set. Note that this option may result in
+	a larger and slower repository; see the discussion in
+	`pack.packSizeLimit`.
 
 --honor-pack-keep::
 	This flag causes an object already in a local pack that
Documentation/git-repack.txt +3 -1
index ef310f362e..24c00c9384 100644
@@ -121,7 +121,9 @@ depth is 4095.
 	If specified, multiple packfiles may be created, which also
 	prevents the creation of a bitmap index.
 	The default is unlimited, unless the config variable
-	`pack.packSizeLimit` is set.
+	`pack.packSizeLimit` is set. Note that this option may result in
+	a larger and slower repository; see the discussion in
+	`pack.packSizeLimit`.
 
 -b::
 --write-bitmap-index::
Signature · 1 line
2.32.0.527.gfd0058899e
#4 lore raw

Re: [PATCH] doc: warn people against --max-pack-size

Documentation/config/pack.txt pack.packSizeLimit:: · line 118

Quoted 2 lines · Jeff King
Documentation/config/pack.txt +2 -0
+The minimum size allowed is limited to 1 MiB. The default is unlimited.
+Common unit suffixes of 'k', 'm', or 'g' are supported.

Do we want to include the workaround of scaling in kibibytes (as originally mentioned by Bagas) for the default as 1024k? This also avoids the easy mistake that the size is in multiples MiB.

Philip

#5 lore raw

Re: [PATCH] doc: warn people against --max-pack-size

Quoted 7 lines · Philip Oakley

Documentation/config/pack.txt pack.packSizeLimit:: · line 118

Quoted 2 lines · Jeff King
Documentation/config/pack.txt +2 -0
+The minimum size allowed is limited to 1 MiB. The default is unlimited.
+Common unit suffixes of 'k', 'm', or 'g' are supported.

Do we want to include the workaround of scaling in kibibytes (as originally mentioned by Bagas) for the default as 1024k? This also avoids the easy mistake that the size is in multiples MiB.

I'm not quite sure what you're asking. If you mean: should we tell people that they can't use "4.9m" and should instead use "5017k" instead, then I don't have a strong opinion.

It might help some people. But OTOH it's not clear to me that this is a common question, so it might clutter up the documentation. Either way, it's orthogonal to the patch in question, and should come on top if somebody cares to work on it.

I'd also be fine with somebody actually implementing fractional unit support (it would probably go into git_parse_signed() and git_parse_unsigned()). It doesn't seem worth the effort to me, but if somebody feels strongly enough to implement it cleanly, I wouldn't say no. :)

-Peff

#6 lore raw

Re: [PATCH] doc: warn people against --max-pack-size

Quoted 11 lines · Jeff King
Quoted 6 lines · Philip Oakley

Documentation/config/pack.txt pack.packSizeLimit:: · line 118

Quoted 2 lines · Jeff King
Documentation/config/pack.txt +2 -0
+The minimum size allowed is limited to 1 MiB. The default is unlimited.
+Common unit suffixes of 'k', 'm', or 'g' are supported.

Do we want to include the workaround of scaling in kibibytes (as originally mentioned by Bagas) for the default as 1024k? This also avoids the easy mistake that the size is in multiples MiB.

I'm not quite sure what you're asking. If you mean: should we tell people that they can't use "4.9m" and should instead use "5017k" instead, then I don't have a strong opinion.

Sorry, I should have included an example based on the patch

Documentation/config/pack.txt pack.packSizeLimit:: · line 118

Quoted 2 lines
Documentation/config/pack.txt +2 -0
+The minimum size allowed is limited to 1 MiB. The default is unlimited.
+Common unit suffixes of 'k', 'm', or 'g' are supported.

e.g.

+The default is unlimited. Common unit suffixes of 'k', 'm', or 'g' are +supported. The minimum size allowed is limited to 1 MiB (1024k).

I swapped the sentence order to allow the scaled example of the minimum to be after the explanation of the suffixes

Quoted 11 lines

It might help some people. But OTOH it's not clear to me that this is a common question, so it might clutter up the documentation. Either way, it's orthogonal to the patch in question, and should come on top if somebody cares to work on it.

I'd also be fine with somebody actually implementing fractional unit support (it would probably go into git_parse_signed() and git_parse_unsigned()). It doesn't seem worth the effort to me, but if somebody feels strongly enough to implement it cleanly, I wouldn't say no. :)

I'd agree about not offering fractional values, but showing the use of smaller units was an easy tweak.

Philip

#7 lore raw

Re: [PATCH] doc: warn people against --max-pack-size

Quoted 14 lines · Philip Oakley
Quoted 3 lines

I'm not quite sure what you're asking. If you mean: should we tell people that they can't use "4.9m" and should instead use "5017k" instead, then I don't have a strong opinion.

Sorry, I should have included an example based on the patch

Documentation/config/pack.txt pack.packSizeLimit:: · line 118

Quoted 2 lines
Documentation/config/pack.txt +2 -0
+The minimum size allowed is limited to 1 MiB. The default is unlimited.
+Common unit suffixes of 'k', 'm', or 'g' are supported.

e.g.

(quoted diff) +2 -0
+The default is unlimited. Common unit suffixes of 'k', 'm', or 'g' are
+supported. The minimum size allowed is limited to 1 MiB (`1024k`). 
Quoted 2 lines

I swapped the sentence order to allow the scaled example of the minimum to be after the explanation of the suffixes

Ah, I see. I don't have a strong feeling either way if you want to prepare a patch.

-Peff