On (29/08/16 07:09), Fabiano Fidêncio wrote:
>Hoiwdy!
>
>
>On Fri, Aug 19, 2016 at 1:08 AM, Fabiano Fidêncio <[email protected]> wrote:
>> This patch is a first attempt to make "make intgcheck" less
>> painful/time consuming than it is now.
>>
>> Although the patch provides a good improvement on having 5 new
>> targets, I know it's still not ideal. The ideal case, IMO, would be
>> being able to select which subset of tests would be run, but that's an
>> improvement that can be done later on.
>>
>> Tips are welcome.
>>
>> Best Regards,
>
>After a few dicussions and some tests done when we met personally, I
>found out that the patch was broken.
>Taking Lukaš suggestion (almost) I've merged the -prepare and
>-configure parts and re-worked the way we get the prefix.
>
>Now it seems to be working!
>
>Best Regards,
>--
>Fabiano Fidêncio

>From a33e1dffd063845e709f5fc1cfec93c330445ab4 Mon Sep 17 00:00:00 2001
>From: =?UTF-8?q?Fabiano=20Fid=C3=AAncio?= <[email protected]>
>Date: Thu, 18 Aug 2016 16:24:17 +0200
>Subject: [PATCH v2] BUILD: Add a few more targets for intg tests
>MIME-Version: 1.0
>Content-Type: text/plain; charset=UTF-8
>Content-Transfer-Encoding: 8bit
>
>Running "make intgcheck" has been proven to be a bit painful (mainly
>when the developer is just writing down a single test case), as it
>cleans up the build directory and fireis a new build before, finally,
>run the tests.
>
>In order to make it a little less painful, let's break the whole
>operation into 4 new targets:
>    intgcheck-{prepare,build,run,clean}.
>
>As expected, "make intgcheck" calls these 4 new operations in the same
>order they were presented, not changing then the current behavior.
>
>Each operation will trigger the previous one in case there is no
>"$$prefix" directory created and the directory is _only_ created in the
>very first operation (intghceck-prepare).
>
>Signed-off-by: Fabiano Fidêncio <[email protected]>
>---
> Makefile.am | 52 ++++++++++++++++++++++++++++++++++++----------------
> 1 file changed, 36 insertions(+), 16 deletions(-)
>
>diff --git a/Makefile.am b/Makefile.am
>index 30d874e..8372b92 100644
>--- a/Makefile.am
>+++ b/Makefile.am
>@@ -3076,30 +3076,50 @@ endif
> # Integration tests #
> #####################
> 
>-intgcheck:
>+intgcheck-prepare:
>       echo "temporarily disabled"
>       set -e; \
>-      rm -Rf intg; \
>-      $(MKDIR_P) intg/bld; \
>-      : Use /hopefully/ short prefix to keep D-Bus socket path short; \
>-      prefix=`mktemp --tmpdir --directory sssd-intg.XXXXXXXX`; \
>-      $(LN_S) "$$prefix" intg/pfx; \
>-      cd intg/bld; \
>+      rm -Rf intg ; \
>+      $(MKDIR_P) intg/bld ; \
>+      : Use /hopefully/ short prefix to keep D-Bus socket path short ; \
>+      prefix=`mktemp --tmpdir --directory sssd-intg.XXXXXXXX` ; \
>+      $(LN_S) "$$prefix" intg/pfx ; \
>+      cd intg/bld ; \
>       $(abs_top_srcdir)/configure \
>-          --prefix="$$prefix" \
>+          --prefix=$$prefix \
>           --with-ldb-lib-dir="$$prefix"/lib/ldb \
>           --enable-intgcheck-reqs \
>           --without-semanage \
>-          $(INTGCHECK_CONFIGURE_FLAGS); \
>-      $(MAKE) $(AM_MAKEFLAGS); \
>-      : Force single-thread install to workaround concurrency issues; \
>-      $(MAKE) $(AM_MAKEFLAGS) -j1 install; \
>-      : Remove .la files from LDB module directory to avoid loader warnings; \
>-      rm "$$prefix"/lib/ldb/*.la; \
>-      $(MAKE) $(AM_MAKEFLAGS) -C src/tests/intg intgcheck-installed; \
>-      cd ../..; \
>+          $(INTGCHECK_CONFIGURE_FLAGS) ; \
>+      cd ../..
>+
>+intgcheck-build:
>+      if [ ! -d intg/pfx ]; then $(MAKE) intgcheck-prepare; fi ; \
>+      prefix=`readlink -e intg/pfx` ; \
>+      cd intg/bld ; \
>+      $(MAKE) $(AM_MAKEFLAGS) ; \
>+      : Force single-thread install to workaround concurrency issues ; \
>+      $(MAKE) $(AM_MAKEFLAGS) -j1 install ; \
>+      : Remove .la files from LDB module directory to avoid loader warnings ; 
>\
>+      rm "$$prefix"/lib/ldb/*.la ; \
>+      cd ../..
>+
>+intgcheck-run:
>+      if [ ! -d intg/pfx ]; then $(MAKE) intgcheck-build; fi ; \
>+      cd intg/bld ; \
>+      $(MAKE) $(AM_MAKEFLAGS) -C src/tests/intg intgcheck-installed ; \
>+      cd ../..
>+
>+intgcheck-clean:
>+      prefix=`readlink -e intg/pfx` ; \
>       rm -Rf "$$prefix" intg
> 
>+intgcheck:
>+      $(MAKE) intgcheck-prepare
>+      $(MAKE) intgcheck-build
I would merge intgcheck-prepare and intgcheck-build bas well.

The reason is quite simple it simplify *my* workflow of
test development.

My current workflow is
* run make intgcheck
* break it before execution of pytest
* change dir to "intg/build"
* finish preparation
  make install
* run tests (or just a particular test)
   make -C src/tests/intg intgcheck-installed INTGCHECK_PYTEST_ARGS="-k 
test_netgroup.py"

(optional)
* clean everything in custom prefix (after breaking stuff there)
  rm -rf ../pfx/*
* prepare stuff for test one more time.
  make install
* run tests one more time
   make -C src/tests/intg intgcheck-installed INTGCHECK_PYTEST_ARGS="-k 
test_netgroup.py"

And now the question is: Do others want to use the same workflow
or we should prefer to run all commands from current
build_dir and not from special build_dir for integration tests.

LS
_______________________________________________
sssd-devel mailing list
[email protected]
https://lists.fedorahosted.org/admin/lists/[email protected]

Reply via email to