Too hard 'make check' yet

[tieugene]

'Make check' still produces too high load at building farms and they stop rpm building because of limits: https://koji.fedoraproject.org/koji/taskinfo?taskID=75900011 Proposition is to split 'make check' into two parts:

1. 'make check' itself that checks functionality correctness (light version)
2. 'make test' (or like this) to run heavy load tests

[tieugene]

Another version - this is bug about x32 that was in about 0.9.7, fixed in about 0.9.16 but returns now

[estraier]

Hi, Could you give me logs around the error?

On Sun, Sep 19, 2021 at 4:14 AM Eugene @.***> wrote:

Another version - this is bug about x32 that was in about 0.9.7, fixed in about 0.9.16 but returns now

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/estraier/tkrzw/issues/23#issuecomment-922359014, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQGJVRA66PAZHVY4VBCDHILUCTQHRANCNFSM5EJPEABA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

[estraier]

I've reproduced the error when I add -m32 to the compiler flags. I'll examine it further.

LD_LIBRARY_PATH=.:/usr/local/lib::. ./tkrzw_dbm_perf wicked --dbm hash --file pos-atom --path casket.tkh \ --block_size 512 --padding --pagecache \ --iter 10000 --threads 4 --size 100 --buckets 100000 --iterator --clear --rebuild --validate Doing: impl=hash num_iterations=10000 value_size=100 num_threads=4 .................................................. (00000500) .................................................. (00001000) .................................................. (00001500) .................................................. (00002000) .................................................. (00002500) .................................................. (00003000) .................................................. (00003500) .................................................. (00004000) .................................................. (00004500) .................................................. (00005000) .................................................. (00005500) .................................................. (00006000) .................................................. (00006500) .................................................. (00007000) .................................................. (00007500) .................................................. (00008000) .................................................. (00008500) .................................................. (00009000) .................................................. (00009500) .................................................. (00010000) Synchronizing: ... done (elapsed=0.000064) Validating records: ... ValidateRecords failed: BROKEN_DATA_ERROR: inconsistent effective data size done (elapsed=0.002479) Done: elapsed_time=0.373422 num_records=1212 qps=107117 mem=4704000 file_size=92016 eff_data_size=38654774211 efficiency=42008753.05% num_buckets=1447 load_factor=0.84

make[1]: [Makefile:291: check-hashdbm-perf] Error 1 make[1]: Leaving directory '/home/mikio/dev/tkrzw' make: [Makefile:143: check] Error 2

On Sun, Sep 19, 2021 at 10:13 AM Mikio Hirabayashi @.***> wrote:

Hi, Could you give me logs around the error?

On Sun, Sep 19, 2021 at 4:14 AM Eugene @.***> wrote:

Another version - this is bug about x32 that was in about 0.9.7, fixed in about 0.9.16 but returns now

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/estraier/tkrzw/issues/23#issuecomment-922359014, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQGJVRA66PAZHVY4VBCDHILUCTQHRANCNFSM5EJPEABA . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub.

[estraier]

I fixed the bug and released 1.0.11.

The new Makefile has the "check-light" target which runs essential and lightweight test cases.

[tieugene]

It works