Can you say what "normal make check" involves and how it failed?
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2020-02-24
OK so the process is (ignoring the -j and CPU_COUNT switches for clarity):
cd $LILYPOND_GIT
mkdir build
./autogen.sh --noconfigure
cd build
../configure --disable-optimising
make
make test-baseline
cd ..
Now I apply patch using:
git apply --index [path to diff from Rietveld]
Now since last week I do this:
cd build
./config.status --recheck
touch config.hh
Then I run
cd ..
./scripts/auxiliar/makelsr.py
This covers any 'snippets/new' type patches
and finally:
make clean
make
make check
make doc
Done.
According to my primitive 'where am I in the process?' logging - which is just a bunch of 'echo' commands to a text file saying when each step completes (and how long it took) this patch fails at some point in the make check stage.
Yet if I start with a clean tree, apply the patch and build LP, test-baselines and all doc with seperate make commands (and nothing else) it all works.
It fails in make check with this (I apologies for the longish output):
This failure is also seen on the other 'make' patches in countdown (issues 5782, 5785 and 5786 for example). Yet all three also pass in the smae way as mentioned above
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
OK so the process is (ignoring the -j and CPU_COUNT switches for clarity):
cd $LILYPOND_GIT
mkdir build
./autogen.sh --noconfigure
cd build
../configure --disable-optimising
make
make test-baseline
cd ..
Now I apply patch using:
git apply --index [path to diff from Rietveld]
Now since last week I do this:
cd build
./config.status --recheck
touch config.hh
Then I run
cd ..
./scripts/auxiliar/makelsr.py
This covers any 'snippets/new' type patches
and finally:
make clean
make
make check
make doc
Done.
According to my primitive 'where am I in the process?' logging - which is just a bunch of 'echo' commands to a text file saying when each step completes (and how long it took) this patch fails at some point in the make check stage.
Yet if I start with a clean tree, apply the patch and build LP, test-baselines and all doc with seperate make commands (and nothing else) it all works.
It fails in make check with this (I apologies for the longish output):
...
Making input/regression/lilypond-book/out-test/texinfo-papersize-docs.pdf < texi
comparing input/regression/out-test-baseline
to input/regression/out-test
2100 signature
1349 log
1348 profile
1 gittxt
4798 total
comparing input/regression/midi/out-test-baseline
to input/regression/midi/out-test
30 signature
39 midi
31 log
30 profile
130 total
comparing input/regression/midi/out-test-baseline/recovered
to input/regression/midi/out-test/recovered
0 total
no source for input/regression/out-test/collated-files.texilog.log
no source for input/regression/out-test/tree.gittxt
no source for input/regression/midi/out-test/dynamic-initial-1.midi
no source for input/regression/midi/out-test/sequence-name-1.midi
no source for input/regression/midi/out-test/sequence-name-2.midi
no source for input/regression/midi/out-test/sequence-name-3.midi
no source for input/regression/midi/out-test/sequence-name-scoping-1.midi
no source for input/regression/midi/out-test/sequence-name-scoping-2.midi
no source for input/regression/midi/out-test/sequence-name-scoping-3.midi
no source for input/regression/midi/out-test/sequence-name-scoping-4.midi
no source for input/regression/midi/out-test/sequence-name-scoping-5.midi
no source for input/regression/midi/out-test/collated-files.texilog.log
Traceback (most recent call last):
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 1602, in <module>
main ()
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 1599, in main
compare_tree_pairs (list(zip(args[0::2], args[1::2])), out, options.threshold)
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 1300, in compare_tree_pairs
data.print_results (threshold)
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 1292, in print_results
self.write_text_result_page ('', threshold)
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 1077, in write_text_result_page
(changed, below, unchanged) = self.thresholded_results (threshold)
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 1055, in thresholded_results
for link in list(self.file_links.values ())]
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 1055, in <listcomp>
for link in list(self.file_links.values ())]
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 396, in distance
self._distance = self.calc_distance ()
File "/home/james/lilypond-git/build/scripts/build/out/output-distance", line 537, in calc_distance
cont0 = self.contents[0].strip();
AttributeError: 'NoneType' object has no attribute 'strip'
/home/james/lilypond-git/build/../GNUmakefile.in:376: recipe for target 'local-check' failed
make: *** [local-check] Error 1</listcomp></module>
This failure is also seen on the other 'make' patches in countdown (issues 5782, 5785 and 5786 for example). Yet all three also pass in the smae way as mentioned above
[issues:#5785] texinfo-rules: generate log files in $(outdir)/
Status: Started
Created: Sat Feb 22, 2020 11:22 PM UTC by Han-Wen Nienhuys
Last Updated: Mon Feb 24, 2020 03:07 PM UTC
Owner: Han-Wen Nienhuys
make clean is not enough. You need make clean test-clean doc-clean in order to be reasonably sure to have a new setup. Of course that comes with the disadvantage that make check and make doc will then do a full build again. But that's just not to be avoided if one wants to check seriously.
Last edit: David Kastrup 2020-02-24
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2020-02-24
OK I can add those. I don't really care how long it takes.
If you are interested each make/make test-baseline/make check takes ~ 2.5 mins and a full make doc takes ~20mins. For about 30 mins in total.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2020-02-26
Patch: review --> new
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2020-02-26
I realised I didn't do a test with the 'normal' patch process, just setting back to new fior now, if it passes, I'll move it to Countdown for Feb 28th
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2020-02-26
Patch: new --> needs_work
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2020-02-26
Even with the make clean commands, this still fails in the same way when apply the patch, then
make clean test-clean doc-clean
make
make check
If I just make, make test-baseline, and doc (with the patch already applied it works).
So I guess patchy merge would fail.
I am not sure how to test this.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Passes make make test-baseline and a full make doc
(again I cannot seem to run a make check on this because after applying the patch, normal make check fails)
Can you say what "normal make check" involves and how it failed?
OK so the process is (ignoring the -j and CPU_COUNT switches for clarity):
Now I apply patch using:
Now since last week I do this:
Then I run
This covers any 'snippets/new' type patches
and finally:
Done.
According to my primitive 'where am I in the process?' logging - which is just a bunch of 'echo' commands to a text file saying when each step completes (and how long it took) this patch fails at some point in the make check stage.
Yet if I start with a clean tree, apply the patch and build LP, test-baselines and all doc with seperate make commands (and nothing else) it all works.
It fails in make check with this (I apologies for the longish output):
This failure is also seen on the other 'make' patches in countdown (issues 5782, 5785 and 5786 for example). Yet all three also pass in the smae way as mentioned above
Thanks. It looks like we are conflating .log files (regression test
output) and pdflatex .log files (uninteresting for testing purposes).
Shall we rename the regtest logs to .testlog or somehting?
On Mon, Feb 24, 2020 at 9:45 PM pkx166h
lilypond-pkx@users.sourceforge.net wrote:
Related
Issues:
#5785make cleanis not enough. You needmake clean test-clean doc-cleanin order to be reasonably sure to have a new setup. Of course that comes with the disadvantage thatmake checkandmake docwill then do a full build again. But that's just not to be avoided if one wants to check seriously.Last edit: David Kastrup 2020-02-24
OK I can add those. I don't really care how long it takes.
If you are interested each make/make test-baseline/make check takes ~ 2.5 mins and a full make doc takes ~20mins. For about 30 mins in total.
I realised I didn't do a test with the 'normal' patch process, just setting back to new fior now, if it passes, I'll move it to Countdown for Feb 28th
Even with the make clean commands, this still fails in the same way when apply the patch, then
make clean test-clean doc-clean
make
make check
If I just make, make test-baseline, and doc (with the patch already applied it works).
So I guess patchy merge would fail.
I am not sure how to test this.
I can run make check now. Try again?
Patch doesn't apply to current master - probably needs a rebase
rebase
https://codereview.appspot.com/557480043
Diff:
Passes make, make check and a full make doc.
Reg test diff attached
diffs are expected.
Patch on countdown for April 9th
Patch counted down - please push.