Re: [PATCH v2 05/11] t0027: make hash size independent'
On 20.08.18 00:10, Eric Sunshine wrote: > On Sun, Aug 19, 2018 at 5:57 PM brian m. carlson > wrote: >> On Sun, Aug 19, 2018 at 04:10:21PM -0400, Eric Sunshine wrote: >>> On Sun, Aug 19, 2018 at 1:54 PM brian m. carlson - tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp" && + tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp+" && >>> >>> My immediate thought upon reading this was whether "+" is valid in >>> Windows filenames. Apparently, it is, but perhaps (if you re-roll) it >>> would make sense to use a character less likely to cause brain >>> hiccups; for instance, "exp0'. >> >> The reason I picked that is that we use it quite a bit in the Makefile, >> so it seemed like an obvious choice for a temporary file name. If you >> feel strongly I can pick something else, but I thought it would be >> reasonably intuitive for other developers. Maybe I was wrong, though. > > I don't feel strongly about it. My brain tripped over it probably > because it's not an idiom in Git tests. In fact, I see just one place > where "+" has been used like this, in t6026. > Probably "tmp" is a better name than "exp+" (Why the '+' ? Is it better that the non-'+' ?) Anyway, If we re-order a little bit, can we use a simple '|' instead ? tr '\015\000abcdef0123456789' QN0 <"$2" | sed -e "s/+/$ZERO_OID/" >"$exp" && tr '\015\000abcdef0123456789' QN0 <"$3" >"$act" &&
Re: [PATCH v2 05/11] t0027: make hash size independent'
On Sun, Aug 19, 2018 at 5:57 PM brian m. carlson wrote: > On Sun, Aug 19, 2018 at 04:10:21PM -0400, Eric Sunshine wrote: > > On Sun, Aug 19, 2018 at 1:54 PM brian m. carlson > > > - tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp" && > > > + tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp+" > > > && > > > > My immediate thought upon reading this was whether "+" is valid in > > Windows filenames. Apparently, it is, but perhaps (if you re-roll) it > > would make sense to use a character less likely to cause brain > > hiccups; for instance, "exp0'. > > The reason I picked that is that we use it quite a bit in the Makefile, > so it seemed like an obvious choice for a temporary file name. If you > feel strongly I can pick something else, but I thought it would be > reasonably intuitive for other developers. Maybe I was wrong, though. I don't feel strongly about it. My brain tripped over it probably because it's not an idiom in Git tests. In fact, I see just one place where "+" has been used like this, in t6026.
Re: [PATCH v2 05/11] t0027: make hash size independent'
On Sun, Aug 19, 2018 at 04:10:21PM -0400, Eric Sunshine wrote:
> On Sun, Aug 19, 2018 at 1:54 PM brian m. carlson
> wrote:
> > We transform various object IDs into all-zero object IDs for comparison.
> > Adjust the length as well so that this works for all hash algorithms.
> >
> > Signed-off-by: brian m. carlson
> > ---
> > diff --git a/t/t0027-auto-crlf.sh b/t/t0027-auto-crlf.sh
> > @@ -15,8 +15,9 @@ compare_ws_file () {
> > - tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp" &&
> > + tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp+" &&
>
> My immediate thought upon reading this was whether "+" is valid in
> Windows filenames. Apparently, it is, but perhaps (if you re-roll) it
> would make sense to use a character less likely to cause brain
> hiccups; for instance, "exp0'.
The reason I picked that is that we use it quite a bit in the Makefile,
so it seemed like an obvious choice for a temporary file name. If you
feel strongly I can pick something else, but I thought it would be
reasonably intuitive for other developers. Maybe I was wrong, though.
--
brian m. carlson: Houston, Texas, US
OpenPGP: https://keybase.io/bk2204
signature.asc
Description: PGP signature
Re: [PATCH v2 05/11] t0027: make hash size independent
On Sun, Aug 19, 2018 at 1:54 PM brian m. carlson
wrote:
> We transform various object IDs into all-zero object IDs for comparison.
> Adjust the length as well so that this works for all hash algorithms.
>
> Signed-off-by: brian m. carlson
> ---
> diff --git a/t/t0027-auto-crlf.sh b/t/t0027-auto-crlf.sh
> @@ -15,8 +15,9 @@ compare_ws_file () {
> - tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp" &&
> + tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp+" &&
My immediate thought upon reading this was whether "+" is valid in
Windows filenames. Apparently, it is, but perhaps (if you re-roll) it
would make sense to use a character less likely to cause brain
hiccups; for instance, "exp0'.
> tr '\015\000abcdef0123456789' QN0 <"$3" >"$act" &&
> + sed -e "s/+/$ZERO_OID/" "$exp+" >"$exp" &&
[PATCH v2 05/11] t0027: make hash size independent
We transform various object IDs into all-zero object IDs for comparison.
Adjust the length as well so that this works for all hash algorithms.
Signed-off-by: brian m. carlson
---
t/t0027-auto-crlf.sh | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/t/t0027-auto-crlf.sh b/t/t0027-auto-crlf.sh
index beb5927f77..e748060489 100755
--- a/t/t0027-auto-crlf.sh
+++ b/t/t0027-auto-crlf.sh
@@ -15,8 +15,9 @@ compare_ws_file () {
pfx=$1
exp=$2.expect
act=$pfx.actual.$3
- tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp" &&
+ tr '\015\000abcdef0123456789' QN0 <"$2" >"$exp+" &&
tr '\015\000abcdef0123456789' QN0 <"$3" >"$act" &&
+ sed -e "s/+/$ZERO_OID/" "$exp+" >"$exp" &&
test_cmp "$exp" "$act" &&
rm "$exp" "$act"
}
