| Message ID | 20170901180340.30009-3-eblake@redhat.com |
|---|---|
| State | New |
| Headers | show |
| Series | Preliminary libqtest cleanups | expand |
On 01.09.2017 20:03, Eric Blake wrote: > It doesn't matter if things are unsorted, but finding stuff in a list > is easier when it is sorted. (Sorted under LC_ALL=C rules, rather than > en_US.UTF-8). > > Signed-off-by: Eric Blake <eblake@redhat.com> > --- > tests/.gitignore | 30 +++++++++++++++--------------- > 1 file changed, 15 insertions(+), 15 deletions(-) Reviewed-by: Thomas Huth <thuth@redhat.com> BTW, did you note the patch "build-sys: generate tests/.gitignore" from Marc-André ? Maybe the list can be generated automatically, too... > diff --git a/tests/.gitignore b/tests/.gitignore > index 64ecd6683b..8d35a58751 100644 > --- a/tests/.gitignore > +++ b/tests/.gitignore > @@ -1,25 +1,27 @@ > +*-test ... but looking at this wildcard, I wonder if we just named most of our tests in a bad way. If we name the files something-test instead of test-something, we maybe do not need the automatic generation from Marc-André (which is kind of ugly because it writes to the source directory - and this should not happen when doing out of tree builds). So what do you think about renaming our tests, so that the above wildcard applies instead? Thomas > atomic_add-bench > benchmark-crypto-cipher > benchmark-crypto-hash > benchmark-crypto-hmac > check-qdict > -check-qnum > check-qjson > check-qlist > check-qnull > -check-qstring > +check-qnum > check-qom-interface > check-qom-proplist > -multiboot/test.out > +check-qstring > multiboot/*.elf > +multiboot/test.out > +qapi-schema/*.test.* > qht-bench > rcutorture > test-aio > test-aio-multithread > test-arm-mptimer > test-base64 > -test-bitops > test-bitcnt > +test-bitops > test-blockjob > test-blockjob-txn > test-bufferiszero > @@ -35,18 +37,19 @@ test-crypto-ivgen > test-crypto-pbkdf > test-crypto-secret > test-crypto-tlscredsx509 > -test-crypto-tlscredsx509-work/ > test-crypto-tlscredsx509-certs/ > +test-crypto-tlscredsx509-work/ > test-crypto-tlssession > -test-crypto-tlssession-work/ > test-crypto-tlssession-client/ > test-crypto-tlssession-server/ > +test-crypto-tlssession-work/ > test-crypto-xts > test-cutils > +test-filter-mirror > +test-filter-redirector > test-hbitmap > test-hmp > test-int128 > -test-iov > test-io-channel-buffer > test-io-channel-command > test-io-channel-command.fifo > @@ -55,27 +58,29 @@ test-io-channel-file.txt > test-io-channel-socket > test-io-channel-tls > test-io-task > +test-iov > test-keyval > test-logging > test-mul64 > +test-netfilter > test-opts-visitor > test-qapi-event.[ch] > test-qapi-types.[ch] > test-qapi-util > test-qapi-visit.[ch] > test-qdev-global-props > -test-qemu-opts > test-qdist > +test-qemu-opts > test-qga > test-qht > test-qht-par > test-qmp-commands > test-qmp-commands.h > test-qmp-event > -test-qobject-input-strict > -test-qobject-input-visitor > test-qmp-introspect.[ch] > test-qmp-marshal.c > +test-qobject-input-strict > +test-qobject-input-visitor > test-qobject-output-visitor > test-rcu-list > test-replication > @@ -92,8 +97,3 @@ test-write-threshold > test-x86-cpuid > test-x86-cpuid-compat > test-xbzrle > -test-netfilter > -test-filter-mirror > -test-filter-redirector > -*-test > -qapi-schema/*.test.* >
Thomas Huth <thuth@redhat.com> writes: > On 01.09.2017 20:03, Eric Blake wrote: >> It doesn't matter if things are unsorted, but finding stuff in a list >> is easier when it is sorted. (Sorted under LC_ALL=C rules, rather than >> en_US.UTF-8). >> >> Signed-off-by: Eric Blake <eblake@redhat.com> >> --- >> tests/.gitignore | 30 +++++++++++++++--------------- >> 1 file changed, 15 insertions(+), 15 deletions(-) > > Reviewed-by: Thomas Huth <thuth@redhat.com> > > BTW, did you note the patch "build-sys: generate tests/.gitignore" from > Marc-André ? Maybe the list can be generated automatically, too... > >> diff --git a/tests/.gitignore b/tests/.gitignore >> index 64ecd6683b..8d35a58751 100644 >> --- a/tests/.gitignore >> +++ b/tests/.gitignore >> @@ -1,25 +1,27 @@ >> +*-test > > ... but looking at this wildcard, I wonder if we just named most of our > tests in a bad way. If we name the files something-test instead of > test-something, we maybe do not need the automatic generation from > Marc-André (which is kind of ugly because it writes to the source > directory - and this should not happen when doing out of tree builds). > > So what do you think about renaming our tests, so that the above > wildcard applies instead? Apropos naming tests: there's an (unspoken) convention to name unit tests check-FOO or test-FOO and tests using libqtest FOO-test, but it's not really honored anymore, probably because people creating tests didn't know about it.
On 05.09.2017 11:53, Markus Armbruster wrote: > Thomas Huth <thuth@redhat.com> writes: > >> On 01.09.2017 20:03, Eric Blake wrote: >>> It doesn't matter if things are unsorted, but finding stuff in a list >>> is easier when it is sorted. (Sorted under LC_ALL=C rules, rather than >>> en_US.UTF-8). >>> >>> Signed-off-by: Eric Blake <eblake@redhat.com> >>> --- >>> tests/.gitignore | 30 +++++++++++++++--------------- >>> 1 file changed, 15 insertions(+), 15 deletions(-) >> >> Reviewed-by: Thomas Huth <thuth@redhat.com> >> >> BTW, did you note the patch "build-sys: generate tests/.gitignore" from >> Marc-André ? Maybe the list can be generated automatically, too... >> >>> diff --git a/tests/.gitignore b/tests/.gitignore >>> index 64ecd6683b..8d35a58751 100644 >>> --- a/tests/.gitignore >>> +++ b/tests/.gitignore >>> @@ -1,25 +1,27 @@ >>> +*-test >> >> ... but looking at this wildcard, I wonder if we just named most of our >> tests in a bad way. If we name the files something-test instead of >> test-something, we maybe do not need the automatic generation from >> Marc-André (which is kind of ugly because it writes to the source >> directory - and this should not happen when doing out of tree builds). >> >> So what do you think about renaming our tests, so that the above >> wildcard applies instead? > > Apropos naming tests: there's an (unspoken) convention to name unit > tests check-FOO or test-FOO and tests using libqtest FOO-test, but it's > not really honored anymore, probably because people creating tests > didn't know about it. Maybe it would help to have an entry for libqtest in MAINTAINERS with a wildcard for tests/*-test.c ? Then we would have at least some kind of documentation for this... (BTW: Any volunteers here for such a maintainer job? ;-)) Thomas
On Tue, Sep 05, 2017 at 11:53:22AM +0200, Markus Armbruster wrote: > Thomas Huth <thuth@redhat.com> writes: > > > On 01.09.2017 20:03, Eric Blake wrote: > >> It doesn't matter if things are unsorted, but finding stuff in a list > >> is easier when it is sorted. (Sorted under LC_ALL=C rules, rather than > >> en_US.UTF-8). > >> > >> Signed-off-by: Eric Blake <eblake@redhat.com> > >> --- > >> tests/.gitignore | 30 +++++++++++++++--------------- > >> 1 file changed, 15 insertions(+), 15 deletions(-) > > > > Reviewed-by: Thomas Huth <thuth@redhat.com> > > > > BTW, did you note the patch "build-sys: generate tests/.gitignore" from > > Marc-André ? Maybe the list can be generated automatically, too... > > > >> diff --git a/tests/.gitignore b/tests/.gitignore > >> index 64ecd6683b..8d35a58751 100644 > >> --- a/tests/.gitignore > >> +++ b/tests/.gitignore > >> @@ -1,25 +1,27 @@ > >> +*-test > > > > ... but looking at this wildcard, I wonder if we just named most of our > > tests in a bad way. If we name the files something-test instead of > > test-something, we maybe do not need the automatic generation from > > Marc-André (which is kind of ugly because it writes to the source > > directory - and this should not happen when doing out of tree builds). > > > > So what do you think about renaming our tests, so that the above > > wildcard applies instead? > > Apropos naming tests: there's an (unspoken) convention to name unit > tests check-FOO or test-FOO and tests using libqtest FOO-test, but it's > not really honored anymore, probably because people creating tests > didn't know about it. $ ls *test.c | wc -l 67 $ ls test*.c | wc -l 73 $ ls check*.c | wc -l 8 Fairly even split between 'test' as a prefix vs suffix. 'check' as a prefix should clearly be killed as a minority pattern. Any appetite for standardizing naming of everything else ? A further idea might be to actually have separate sub-directories for unit tests vs qtests. eg go for tests +- unit +- qtest +- iotests Regards, Daniel
On 05.09.2017 12:10, Daniel P. Berrange wrote: > On Tue, Sep 05, 2017 at 11:53:22AM +0200, Markus Armbruster wrote: >> Thomas Huth <thuth@redhat.com> writes: >> >>> On 01.09.2017 20:03, Eric Blake wrote: >>>> It doesn't matter if things are unsorted, but finding stuff in a list >>>> is easier when it is sorted. (Sorted under LC_ALL=C rules, rather than >>>> en_US.UTF-8). >>>> >>>> Signed-off-by: Eric Blake <eblake@redhat.com> >>>> --- >>>> tests/.gitignore | 30 +++++++++++++++--------------- >>>> 1 file changed, 15 insertions(+), 15 deletions(-) >>> >>> Reviewed-by: Thomas Huth <thuth@redhat.com> >>> >>> BTW, did you note the patch "build-sys: generate tests/.gitignore" from >>> Marc-André ? Maybe the list can be generated automatically, too... >>> >>>> diff --git a/tests/.gitignore b/tests/.gitignore >>>> index 64ecd6683b..8d35a58751 100644 >>>> --- a/tests/.gitignore >>>> +++ b/tests/.gitignore >>>> @@ -1,25 +1,27 @@ >>>> +*-test >>> >>> ... but looking at this wildcard, I wonder if we just named most of our >>> tests in a bad way. If we name the files something-test instead of >>> test-something, we maybe do not need the automatic generation from >>> Marc-André (which is kind of ugly because it writes to the source >>> directory - and this should not happen when doing out of tree builds). >>> >>> So what do you think about renaming our tests, so that the above >>> wildcard applies instead? >> >> Apropos naming tests: there's an (unspoken) convention to name unit >> tests check-FOO or test-FOO and tests using libqtest FOO-test, but it's >> not really honored anymore, probably because people creating tests >> didn't know about it. > > $ ls *test.c | wc -l > 67 > $ ls test*.c | wc -l > 73 > $ ls check*.c | wc -l > 8 > > Fairly even split between 'test' as a prefix vs suffix. 'check' as a > prefix should clearly be killed as a minority pattern. Any appetite > for standardizing naming of everything else ? > > A further idea might be to actually have separate sub-directories for > unit tests vs qtests. eg go for > > tests > +- unit > +- qtest > +- iotests That sounds like a good idea to me, too. Thomas
On 09/05/2017 05:10 AM, Daniel P. Berrange wrote: > Fairly even split between 'test' as a prefix vs suffix. 'check' as a > prefix should clearly be killed as a minority pattern. Any appetite > for standardizing naming of everything else ? > > A further idea might be to actually have separate sub-directories for > unit tests vs qtests. eg go for > > tests > +- unit > +- qtest > +- iotests I like the idea, but don't know if I want to be the one tackling it...
diff --git a/tests/.gitignore b/tests/.gitignore index 64ecd6683b..8d35a58751 100644 --- a/tests/.gitignore +++ b/tests/.gitignore @@ -1,25 +1,27 @@ +*-test atomic_add-bench benchmark-crypto-cipher benchmark-crypto-hash benchmark-crypto-hmac check-qdict -check-qnum check-qjson check-qlist check-qnull -check-qstring +check-qnum check-qom-interface check-qom-proplist -multiboot/test.out +check-qstring multiboot/*.elf +multiboot/test.out +qapi-schema/*.test.* qht-bench rcutorture test-aio test-aio-multithread test-arm-mptimer test-base64 -test-bitops test-bitcnt +test-bitops test-blockjob test-blockjob-txn test-bufferiszero @@ -35,18 +37,19 @@ test-crypto-ivgen test-crypto-pbkdf test-crypto-secret test-crypto-tlscredsx509 -test-crypto-tlscredsx509-work/ test-crypto-tlscredsx509-certs/ +test-crypto-tlscredsx509-work/ test-crypto-tlssession -test-crypto-tlssession-work/ test-crypto-tlssession-client/ test-crypto-tlssession-server/ +test-crypto-tlssession-work/ test-crypto-xts test-cutils +test-filter-mirror +test-filter-redirector test-hbitmap test-hmp test-int128 -test-iov test-io-channel-buffer test-io-channel-command test-io-channel-command.fifo @@ -55,27 +58,29 @@ test-io-channel-file.txt test-io-channel-socket test-io-channel-tls test-io-task +test-iov test-keyval test-logging test-mul64 +test-netfilter test-opts-visitor test-qapi-event.[ch] test-qapi-types.[ch] test-qapi-util test-qapi-visit.[ch] test-qdev-global-props -test-qemu-opts test-qdist +test-qemu-opts test-qga test-qht test-qht-par test-qmp-commands test-qmp-commands.h test-qmp-event -test-qobject-input-strict -test-qobject-input-visitor test-qmp-introspect.[ch] test-qmp-marshal.c +test-qobject-input-strict +test-qobject-input-visitor test-qobject-output-visitor test-rcu-list test-replication @@ -92,8 +97,3 @@ test-write-threshold test-x86-cpuid test-x86-cpuid-compat test-xbzrle -test-netfilter -test-filter-mirror -test-filter-redirector -*-test -qapi-schema/*.test.*
It doesn't matter if things are unsorted, but finding stuff in a list is easier when it is sorted. (Sorted under LC_ALL=C rules, rather than en_US.UTF-8). Signed-off-by: Eric Blake <eblake@redhat.com> --- tests/.gitignore | 30 +++++++++++++++--------------- 1 file changed, 15 insertions(+), 15 deletions(-)