[C-prog-lang-l] phase 1 testing

[Jan Pechanec]

On Tue, 21 Apr 2020, Jan Pechanec wrote:

>On Tue, 21 Apr 2020, Tomáš Sláma wrote:
>
>> Zdravím
>>
>> Raději píšu pouze vám, abych nespamoval skupinu :).
>
>I'm adding the mailing list back, it's really the purpose of the 
>mailing conference to have such discussions.  Please feel free to use 
>it.
>
>>> s/MYGUTAR/MYGNUTAR/
>>
>> Jej, omlouvám se, toho jsem si nevšiml :/.
>>
>> Chybu jsem napravil a testů mi nyní padá 8:
>
>I can see the same thing on a lab machine in the MS building.
>
>Sorry about that.  I will take a look at it, hopefully within a few 
>days.
>
>My macOS machines have GNU tar 1.32 which is the same version as on 
>that Linux box I just used and I don't see right now what is the 
>problem.
>
>I only used my macOS machines when writing the tests and forgot to 
>run it on a Linux distro as well.  Rightfully punished...
>
>So, during your development, your implementation should fail in the 
>same way as GNU tar on a Linux box until I figure it out and fix it.

I believe I fixed the tests.  I can see all tests now pass both on my 
macOS and on a Linux machine in the lab.

There were two issues:

- sort(1) may behave differently under different locale settings.  
  LC_ALL=C is a good way to ensure identical behavior.

- /dev/random may not give you all the data you need as the system may 
  not have enough entropy at the moment.  /dev/urandom should always 
  give you what you ask for but the randomness quality may be worse.

Jan.

-- 
Jan Pechanec <jp (at) devnull (dot) cz>
http://www.devnull.cz