Fix expected values of setup tests on Windows
A message that is not in the archive (sent off-list or not yet received): [email protected]
2010/12/31 Johannes Sixt <[email protected]>:
Quoted 4 lines
On Windows, bash stores absolute path names in shell variables in POSIX format that begins with a slash, rather than in drive-letter format; such a value is converted to the latter format when it is passed to a non-MSYS program such as git.
Hmm.. from test-lib.sh:
TEST_DIRECTORY=$(pwd)
test="trash directory.$(basename "$0" .sh)"
TRASH_DIRECTORY="$TEST_DIRECTORY/$test"
I'm just curious how these lines make $TRASH_DIRECTORY in POSIX format, while
here=$(pwd)
in your patch does not. Does bash auto convert value in TRASH_DIRECTORY="$TE..." line?
Signature · 1 line
Duy
Quoted 12 lines · Nguyen Thai Ngoc Duy
2010/12/31 Johannes Sixt <[email protected]>:
Quoted 4 lines
On Windows, bash stores absolute path names in shell variables in POSIX format that begins with a slash, rather than in drive-letter format; such a value is converted to the latter format when it is passed to a non-MSYS program such as git.
Hmm.. from
test-lib.sh:TEST_DIRECTORY=$(pwd)
test="trash directory.$(basename "$0" .sh)"
TRASH_DIRECTORY="$TEST_DIRECTORY/$test"
Quoted 7 lines
I'm just curious how these lines make $TRASH_DIRECTORY in POSIX format, while
here=$(pwd)
in your patch does not. Does bash auto convert value in TRASH_DIRECTORY="$TE..." line?
No. When this line is executed:
TEST_DIRECTORY=$(pwd)
$(pwd) still has its default behavior to return the POSIX style path. pwd is redefined to pwd -W only later.
I'm hesitant to redefine pwd earlier in test-lib.sh, though, because we would have to audit all uses of TEST_DIRECTORY for whether POSIX style paths or drive-letter paths are needed.
-- Hannes
Quoted 11 lines · Johannes Sixt
Quoted 2 lines · Nguyen Thai Ngoc Duy
in your patch does not. Does bash auto convert value in TRASH_DIRECTORY="$TE..." line?
No. When this line is executed:
TEST_DIRECTORY=$(pwd)
$(pwd) still has its default behavior to return the POSIX style path. pwd is redefined to pwd -W only later.
Would it make sense to change it to
TEST_DIRECTORY=$PWD
for clarity and robustness against code movement, then?
Quoted 9 lines · Jonathan Nieder
Quoted 8 lines · Johannes Sixt
Quoted 2 lines · Nguyen Thai Ngoc Duy
in your patch does not. Does bash auto convert value in TRASH_DIRECTORY="$TE..." line?
No. When this line is executed:
TEST_DIRECTORY=$(pwd)
Quoted 8 lines
Quoted 2 lines
$(pwd) still has its default behavior to return the POSIX style path. pwd is redefined to pwd -W only later.
Would it make sense to change it to
TEST_DIRECTORY=$PWD
for clarity and robustness against code movement, then?
Yes, that would make sense.
-- Hannes
Quoted 13 lines · Johannes Sixt
Quoted 12 lines · Jonathan Nieder
Quoted 10 lines · Johannes Sixt
Quoted 2 lines · Nguyen Thai Ngoc Duy
in your patch does not. Does bash auto convert value in TRASH_DIRECTORY="$TE..." line?
No. When this line is executed:
TEST_DIRECTORY=$(pwd)
$(pwd) still has its default behavior to return the POSIX style path. pwd is redefined to pwd -W only later.
Quoted 7 lines
Quoted 5 lines
Would it make sense to change it to
TEST_DIRECTORY=$PWD
for clarity and robustness against code movement, then?
Yes, that would make sense.
It will be very much appreciated to add a few sentences to clarify this to "Do's and don'ts" section of t/README if you are re-rolling this. Thanks.
A message that is not in the archive (sent off-list or not yet received): [email protected]
PATCH 2/2 t/README: hint about using $(pwd) rather than $PWD in tests
From: Johannes Sixt <[email protected]>
This adds just a "do it this way" instruction without a lot of explanation, because the details are too complex to be explained at this point.
--- t/README | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-)
t/README +6 -0
@@ -283,6 +283,12 @@ Do: Tests that are likely to smoke out future regressions are better than tests that just inflate the coverage metrics. + - When a test checks for an absolute path that a git command generated, + construct the expected value using $(pwd) rather than $PWD, + $TEST_DIRECTORY, or $TRASH_DIRECTORY. It makes a difference on + Windows, where the shell (MSYS bash) mangles absolute path names. + For details, see the commit message of 4114156ae9. + Don't: - exit() within a <script> part.
Signature · 1 line
1.7.4.rc1.1258.g84aa
Re: [PATCH 2/2] t/README: hint about using $(pwd) rather than $PWD in tests
Quoted 2 lines · Johannes Sixt
This adds just a "do it this way" instruction without a lot of explanation, because the details are too complex to be explained at this point.
Thanks, looks very useful.
t/README Do: · line 291
Quoted 12 lines
t/README +6
-0@@ -283,6 +283,12 @@ Do: Tests that are likely to smoke out future regressions are better than tests that just inflate the coverage metrics. + - When a test checks for an absolute path that a git command generated, + construct the expected value using $(pwd) rather than $PWD, + $TEST_DIRECTORY, or $TRASH_DIRECTORY. It makes a difference on + Windows, where the shell (MSYS bash) mangles absolute path names. + For details, see the commit message of 4114156ae9. +
Perhaps it is also worth explaining the cases where $PWD is needed?
By contrast, when a passing a path to git or constructing a URL, use $PWD. It makes a difference on Windows, where
- $(pwd) is a Windows-style path such as git might output, and
- $PWD is a Unix-style path that the shell (MSYS bash) will mangle before passing to native apps like git.
Re: [PATCH 2/2] t/README: hint about using $(pwd) rather than $PWD in tests
Am 1/11/2011 8:54, schrieb Jonathan Nieder:
Quoted 4 lines
Perhaps it is also worth explaining the cases where $PWD is needed?
By contrast, when a passing a path to git or constructing a URL, use $PWD.
The first part of the "or" is not true: you can pass the result of $(pwd) to a command, and it means the same as $PWD; I would even recommend against $PWD so that a reader does not have to wonder "why pass $PWD, but check for $(pwd)?"
The second part I don't know whether it is true: I haven't noticed a pattern where people did it the wrong way, therefore, I'don't even know whether $PWD is really *always* required. Do *you* know?
Quoted 5 lines
It makes a difference on Windows, where
- $(pwd) is a Windows-style path such as git might output, and
- $PWD is a Unix-style path that the shell (MSYS bash) will mangle before passing to native apps like git.
This information is already included by reference to 4114156ae9.
-- Hannes
Re: [PATCH 2/2] t/README: hint about using $(pwd) rather than $PWD in tests
Quoted 11 lines · Johannes Sixt
Am 1/11/2011 8:54, schrieb Jonathan Nieder:
Quoted 4 lines
Perhaps it is also worth explaining the cases where $PWD is needed?
By contrast, when a passing a path to git or constructing a URL, use $PWD.
The first part of the "or" is not true: you can pass the result of $(pwd) to a command, and it means the same as $PWD; I would even recommend against $PWD so that a reader does not have to wonder "why pass $PWD, but check for $(pwd)?"
I _think_ that passing $PWD always gives the right result. By contrast, constructions like
PATH=$(pwd)/bin:$PATH
break iirc.
I suspect that the reader will end up wondering "why does this have to be so complicated" no matter what.
Quoted 3 lines
The second part I don't know whether it is true: I haven't noticed a pattern where people did it the wrong way, therefore, I'don't even know whether $PWD is really *always* required. Do *you* know?
24f1136 is one example. I don't know of any utility that treats
file://c:/foo/bar/baz
as a URL representing a resource on localhost (and msys bash has no rewriting rule for it), so in that particular case (file://$directory), $PWD really does seem to be always required.
Quoted 7 lines
Quoted 5 lines
It makes a difference on Windows, where
- $(pwd) is a Windows-style path such as git might output, and
- $PWD is a Unix-style path that the shell (MSYS bash) will mangle before passing to native apps like git.This information is already included by reference to
4114156ae9.
... but if we can summarize it nicely, we can save the reader a step, no?
Anyway, what you have already written is useful; clearing up these details would just be icing on the top.
Re: [PATCH 2/2] t/README: hint about using $(pwd) rather than $PWD in tests
Am 1/11/2011 9:37, schrieb Jonathan Nieder:
Quoted 2 lines
I suspect that the reader will end up wondering "why does this have to be so complicated" no matter what.
Unfortunately, yes. Therefore, I'd like to keep the paragraph minimal, focused on how expected values should be constructed, which is where errors will happen primarily.
Quoted 10 lines
Quoted 7 lines
Quoted 5 lines
It makes a difference on Windows, where
- $(pwd) is a Windows-style path such as git might output, and
- $PWD is a Unix-style path that the shell (MSYS bash) will mangle before passing to native apps like git.This information is already included by reference to
4114156ae9.... but if we can summarize it nicely, we can save the reader a step, no?
I don't think so: it's not complete enough. Readers will ask: "So what?" Digging archives or a three paragraph follow-up explanation on the ML will be required anyway.
Quoted 2 lines
Anyway, what you have already written is useful; clearing up these details would just be icing on the top.
OK, thanks for a review.
-- Hannes
Re: [PATCH 2/2] t/README: hint about using $(pwd) rather than $PWD in tests
Quoted 7 lines · Johannes Sixt
Am 1/11/2011 9:37, schrieb Jonathan Nieder:
Quoted 2 lines
I suspect that the reader will end up wondering "why does this have to be so complicated" no matter what.
Unfortunately, yes. Therefore, I'd like to keep the paragraph minimal, focused on how expected values should be constructed, which is where errors will happen primarily.
I really was hoping for something simple like "never use $PWD" myself, though X-<.
Anyway, will queue for 1.7.4. Thanks.
Quoted 5 lines · Johannes Sixt
Quoted 2 lines
in your patch does not. Does bash auto convert value in TRASH_DIRECTORY="$TE..." line?
No. When this line is executed:
Quoted 8 lines
TEST_DIRECTORY=$(pwd)
$(pwd) still has its default behavior to return the POSIX style path. pwd is redefined to pwd -W only later.
I'm hesitant to redefine pwd earlier in
test-lib.sh, though, because we would have to audit all uses ofTEST_DIRECTORYfor whether POSIX style paths or drive-letter paths are needed.
Ah I missed that. Thanks. Ack from me.
Signature · 1 line
Duy