pre-commit.sample: don't use [...] around a tr range
PATCH 1/1 pre-commit.sample: don't use [...] around a tr range
Using square brackets in tr ranges is risky.
From the documentation of GNU tr:
Ranges
The notation M-N' expands to all of the characters from M through N, in ascending order. M should collate before N; if it doesn't, an error results. As an example, 0-9' is the same as `0123456789'.
GNU `tr' does not support the System V syntax that uses square brackets to enclose ranges. Translations specified in that format sometimes work as expected, since the brackets are often transliterated to themselves. However, they should be avoided because they sometimes behave unexpectedly. For example, `tr -d '[0-9]'' deletes brackets as well as digits.
However, if the use of [] is deliberate, because git still cares about portability to ancient SYSV versions of tr that require that notation, then let me know and I'll undo that part of the change and add a comment to that effect.
Quoted 1 line
From
cd3c17c6a48c67d2a598001272f3283714b1df19Mon Sep 17 00:00:00 2001
From: Jim Meyering <[email protected]>
Date: Mon, 21 Sep 2009 10:58:02 +0200
Subject: [PATCH] pre-commit.sample: don't use [...] around a tr range
Using square brackets around a tr range is especially risky when using tr's -d (delete) option. In this case, the brackets happen to be harmless, since the two bracket bytes happen to be included in the desired range. Correct spelling and grammar.
--- templates/hooks--pre-commit.sample | 10 +++++----- 1 files changed, 5 insertions(+), 5 deletions(-)
templates/hooks--pre-commit.sample +5 -5
@@ -15,14 +15,14 @@ allownonascii=$(git config hooks.allownonascii) # printable range starts at the space character and ends with tilde. if [ "$allownonascii" != "true" ] && test "$(git diff --cached --name-only --diff-filter=A -z | - LC_ALL=C tr -d '[ -~]\0')" + LC_ALL=C tr -d ' -~\0')" then - echo "Error: Attempt to add a non-ascii filename." + echo "Error: Attempt to add a non-ascii file name." echo - echo "This can cause problems if you want to work together" - echo "with people on other platforms than you." + echo "This can cause problems if you want to work" + echo "with people on other platforms." echo - echo "To be portable it is adviseable to rename the file ..." + echo "To be portable it is advisable to rename the file ..." echo echo "If you know what you are doing you can disable this" echo "check using:"
--
1.6.5.rc1.214.g13c5a
Quoted 4 lines · Jim Meyering
However, if the use of [] is deliberate, because git still cares about portability to ancient SYSV versions of tr that require that notation, then let me know and I'll undo that part of the change and add a comment to that effect.
We have (had?) people trying to support Git on HP-UX and SunOS. Do these count?
Quoted 8 lines · Alex Riesen
Quoted 4 lines · Jim Meyering
However, if the use of [] is deliberate, because git still cares about portability to ancient SYSV versions of tr that require that notation, then let me know and I'll undo that part of the change and add a comment to that effect.
We have (had?) people trying to support Git on HP-UX and SunOS. Do these count?
I had my doubts, but have just confirmed that Solaris 10's /usr/bin/tr is still doing it the SYSV way:
$ echo foo | LC_ALL=C /usr/bin/tr a-z A-Z
fooThere, you have to use /usr/xpg4/bin/tr to get the expected behavior:
$ echo foo | LC_ALL=C /usr/xpg4/bin/tr a-z A-Z
FOOSo you're right. Thanks!
Here's an updated patch:
Quoted 1 line
From
4048a5b393fa5f35dfe8157e98a6ea475b0efb2dMon Sep 17 00:00:00 2001
From: Jim Meyering <[email protected]>
Date: Mon, 21 Sep 2009 10:58:02 +0200
Subject: [PATCH] pre-commit.sample: add comment re tr portability; fix grammar
Add a comment explaining why square brackets around a tr range are not only ok, but actually required in this case. Correct spelling and grammar.
--- templates/hooks--pre-commit.sample | 11 +++++++---- 1 files changed, 7 insertions(+), 4 deletions(-)
templates/hooks--pre-commit.sample +7 -4
@@ -14,15 +14,18 @@ allownonascii=$(git config hooks.allownonascii) # them from being added to the repository. We exploit the fact that the # printable range starts at the space character and ends with tilde. if [ "$allownonascii" != "true" ] && + # Note that the use of brackets around a tr range is ok here, (it's + # even required, for portability to Solaris 10's /usr/bin/tr), since + # the square bracket bytes happen to fall in the designated range. test "$(git diff --cached --name-only --diff-filter=A -z | LC_ALL=C tr -d '[ -~]\0')" then - echo "Error: Attempt to add a non-ascii filename." + echo "Error: Attempt to add a non-ascii file name." echo - echo "This can cause problems if you want to work together" - echo "with people on other platforms than you." + echo "This can cause problems if you want to work" + echo "with people on other platforms." echo - echo "To be portable it is adviseable to rename the file ..." + echo "To be portable it is advisable to rename the file ..." echo echo "If you know what you are doing you can disable this" echo "check using:"
--
1.6.5.rc1.214.g13c5a
Quoted 9 lines · Jim Meyering
Quoted 2 lines
We have (had?) people trying to support Git on HP-UX and SunOS. Do these count?
I had my doubts, but have just confirmed that Solaris 10's
/usr/bin/tris still doing it the SYSV way:$ echo foo | LC_ALL=C /usr/bin/tr a-z A-Z foo
Quoted 6 lines
There, you have to use
/usr/xpg4/bin/trto get the expected behavior:$ echo foo | LC_ALL=C /usr/xpg4/bin/tr a-z A-Z FOOSo you're right. Thanks!
See:
http://article.gmane.org/gmane.comp.version-control.git/76991
templates/hooks--pre-commit.sample allownonascii=$(git config hooks.allownonascii) · line 21
Quoted 5 lines
templates/hooks--pre-commit.sample +3
-0+ # Note that the use of brackets around a tr range is ok here, (it's + # even required, for portability to Solaris 10's /usr/bin/tr), since + # the square bracket bytes happen to fall in the designated range. test "$(git diff --cached --name-only --diff-filter=A -z | LC_ALL=C tr -d '[ -~]\0')"
Does this work on non-bracket systems? I would think that enumerating the sequence would be the most portable thing.
-Peff
Quoted 22 lines · Jeff King
Quoted 9 lines · Jim Meyering
Quoted 2 lines
We have (had?) people trying to support Git on HP-UX and SunOS. Do these count?
I had my doubts, but have just confirmed that Solaris 10's
/usr/bin/tris still doing it the SYSV way:$ echo foo | LC_ALL=C /usr/bin/tr a-z A-Z fooQuoted 6 lines
There, you have to use
/usr/xpg4/bin/trto get the expected behavior:$ echo foo | LC_ALL=C /usr/xpg4/bin/tr a-z A-Z FOOSo you're right. Thanks!
See:
http://article.gmane.org/gmane.comp.version-control.git/76991
Quoted 7 lines
Review
Quoted 5 lines
(quoted diff) +3
-0+ # Note that the use of brackets around a tr range is ok here, (it's + # even required, for portability to Solaris 10's /usr/bin/tr), since + # the square bracket bytes happen to fall in the designated range. test "$(git diff --cached --name-only --diff-filter=A -z | LC_ALL=C tr -d '[ -~]\0')"Does this work on non-bracket systems?
Yes, since [] happen to fall in the range.
Quoted 2 lines
I would think that enumerating
the sequence would be the most portable thing.
Enumerating is more portable, at the expense of readability and maintainability.
In case you want to go that route, here's one more:
(note that this, like the original range, treats TAB as nonportable)
Quoted 1 line
From
40a368a7bcf0ac6524bbe36ba3bfdaa0711897b8Mon Sep 17 00:00:00 2001
From: Jim Meyering <[email protected]>
Date: Mon, 21 Sep 2009 10:58:02 +0200
Subject: [PATCH] pre-commit.sample: use tr more portability; fix grammar
Avoid tr's M-N range notation altogether. Instead, enumerate ascii bytes 32..126. Correct spelling and grammar.
--- templates/hooks--pre-commit.sample | 19 +++++++++++++------ 1 files changed, 13 insertions(+), 6 deletions(-)
templates/hooks--pre-commit.sample +13 -6
@@ -10,19 +10,26 @@ # If you want to allow non-ascii filenames set this variable to true. allownonascii=$(git config hooks.allownonascii) +printables=' !"#$%&'\''()*+,-./0123456789:;<=>?@' +printables="$printables"'ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`' +printables="$printables"'abcdefghijklmnopqrstuvwxyz{|}~' + # Cross platform projects tend to avoid non-ascii filenames; prevent # them from being added to the repository. We exploit the fact that the # printable range starts at the space character and ends with tilde. if [ "$allownonascii" != "true" ] && - test "$(git diff --cached --name-only --diff-filter=A -z | - LC_ALL=C tr -d '[ -~]\0')" + # Filter out printable ascii bytes, and NUL. + # If anything remains, you lose. + rem=$(git diff --cached --name-only --diff-filter=A -z | + LC_ALL=C tr -d "$printables"'\0') + test -n "$rem" then - echo "Error: Attempt to add a non-ascii filename." + echo "Error: Attempt to add a non-ascii file name." echo - echo "This can cause problems if you want to work together" - echo "with people on other platforms than you." + echo "This can cause problems if you want to work" + echo "with people on other platforms." echo - echo "To be portable it is adviseable to rename the file ..." + echo "To be portable it is advisable to rename the file ..." echo echo "If you know what you are doing you can disable this" echo "check using:"
--
1.6.5.rc1.214.g13c5a
Quoted 5 lines · Jim Meyering
Quoted 4 lines
Review
Quoted 2 lines
(quoted diff) +0
-0test "$(git diff --cached --name-only --diff-filter=A -z | LC_ALL=C tr -d '[ -~]\0')"Does this work on non-bracket systems?
Quoted 8 lines
Yes, since [] happen to fall in the range.
Quoted 2 lines
I would think that enumerating
the sequence would be the most portable thing.Enumerating is more portable, at the expense of readability and maintainability.
In case you want to go that route, here's one more:
Agreed. If we can avoid enumeration, we should. If your original is portable, then I think it is preferable. Thanks for looking into it.
-Peff
Quoted 10 lines · Jeff King
Quoted 8 lines · Jim Meyering
Quoted 2 lines
We have (had?) people trying to support Git on HP-UX and SunOS. Do these count?
I had my doubts, but have just confirmed that Solaris 10's
/usr/bin/tris still doing it the SYSV way:$ echo foo | LC_ALL=C /usr/bin/tr a-z A-Z foo
Quoted 6 lines
Quoted 6 lines
There, you have to use
/usr/xpg4/bin/trto get the expected behavior:$ echo foo | LC_ALL=C /usr/xpg4/bin/tr a-z A-Z FOOSo you're right. Thanks!
By the way, modern git inserts /usr/xpg4/bin into PATH before /usr/bin on Solaris. So /usr/xpg4/bin/tr should always be used on that platform.
-brandon
Quoted 8 lines · Brandon Casey
Quoted 8 lines
Quoted 8 lines
I had my doubts, but have just confirmed that Solaris 10's
/usr/bin/tris still doing it the SYSV way:$ echo foo | LC_ALL=C /usr/bin/tr a-z A-Z fooThere, you have to use
/usr/xpg4/bin/trto get the expected behavior:
Quoted 7 lines
Quoted 4 lines
Quoted 4 lines
$ echo foo | LC_ALL=C /usr/xpg4/bin/tr a-z A-Z FOOSo you're right. Thanks!
By the way, modern git inserts
/usr/xpg4/bininto PATH before/usr/binon Solaris. So/usr/xpg4/bin/trshould always be used on that platform.
Ah, I forgot about that patch. So it may not matter on Solaris. Are there other platforms which want the brackets? Maybe HP-UX, but I no longer have access to a box and I've used deep hypnosis to suppress any memories of ever having used it.
Anyway, I think Jim's patch is the right fix if there are still platforms that want the brackets.
-Peff
Quoted 16 lines · Jim Meyering
Quoted 8 lines · Alex Riesen
Quoted 4 lines · Jim Meyering
However, if the use of [] is deliberate, because git still cares about portability to ancient SYSV versions of tr that require that notation, then let me know and I'll undo that part of the change and add a comment to that effect.
We have (had?) people trying to support Git on HP-UX and SunOS. Do these count?
I had my doubts, but have just confirmed that Solaris 10's
/usr/bin/tris still doing it the SYSV way:$ echo foo | LC_ALL=C /usr/bin/tr a-z A-Z foo
Quoted 6 lines
There, you have to use
/usr/xpg4/bin/trto get the expected behavior:$ echo foo | LC_ALL=C /usr/xpg4/bin/tr a-z A-Z FOOSo you're right. ...
Sorry about that. I even made a simple tr once for Git tests exactly because of compatibility problems (look for "poor man tr" in the list archives, if needed).