Bug#960681: Bug#513967:l icensecheck: fails to detect license at end (option --tail is broken)

[Jonas Smedegaard]

Quoting Dominique Dumont (2021-11-21 18:32:31)
> On Saturday, 20 November 2021 11:15:59 CET Jonas Smedegaard wrote:
> > I would appreciate some numbers about actual slowdown.
> 
> Fair enough.
> 
> Here are some measurements where the cell content is the "real" time given by time command.
> 
> This table is to be viewed with a monospace font.
> 
> licensecheck command is:
> ┌────
> │ licensecheck --lines 0 --encoding utf8 --copyright --machine --shortname-scheme=debian,spdx --recursive .
> └────
> 
> This is also the command used internally by cme.
> 
> ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
>  package  plain      cme with   licensecheck  licensecheck 
>           cme        lines=0                  with lines=0 
> ───────────────────────────────────────────────────────────
>  pan      0m2.694s   0m6.553s   0m4.571s      0m9.303s     
>  moarvm   0m3.768s   0m41.772s  0m3.900s      0m40.274s    
>  nqp      0m3.057s   0m3.635s   0m3.682s      0m9.955s     
>  rakudo   0m3.448s   0m9.784s   0m11.358s     0m17.517s    
>  systemd  4m30.489s  4m59.546s  4m31.644s     5m2.661s     
> ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
> 
> 
> The result is surprising as using --lines 0 can be lead to similar time or 10 times longer...

Thanks.

Takes longer indeed.

Only systemd is relatively large, though, with only little slowdown.

Only moarvm has extreme slowdown, which is vastly reduced by skipping a 
few large test files codifying random numbers, by adding this option:

  --ignore '3rdparty/ryu/third_party/double-conversion/test/cctest/gay-.*cc'

Also, systemd seems to contain tidy SPDX annotations, which is detected 
even scanning only top 20 lines of code.

A project I know is both relatively large and more messy is emscripten.

On my system, scanning emscripten takes 1m21,149s using this command:

  licensecheck --check '.*' --recursive --copyright --deb-machine --lines 0 --merge-licenses --ignore '.*\.(png|mdl|wav|ogg|mp3|dds)$' -- *

Dropping "--lines 0" it takes only 0m35,730s

But comparing the output shows quite a few differences - e.g. the 
FreeType Project is not detected as copyright holder.

I find it relevant to take loss of accuracy into account.


 - Jonas

-- 
 * Jonas Smedegaard - idealist & Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: signature
URL: <http://alioth-lists.debian.net/pipermail/pkg-perl-maintainers/attachments/20211121/74f7f730/attachment.sig>