cmpilato@collab.net wrote:
>brane@tigris.org writes:
>
>
>
>>Author: brane
>>Date: 2002-12-05 13:24:07 -0600 (Thu, 05 Dec 2002)
>>New Revision: 4013
>>
>>Added:
>> trunk/subversion/tests/clients/cmdline/theta.bin
>>Removed:
>> trunk/subversion/tests/clients/cmdline/theta.png
>>Modified:
>> trunk/subversion/tests/clients/cmdline/update_tests.py
>>Log:
>>Fix update_tests.py to pass on Windows. The theta.png file was a very
>>beautiful image of a theta, but sadly it contaied ^Z characters. Those
>>are end-of-file markers on Windows, and the tests read the all files
>>in text mode...well, results were predictable.
>>
>>
>
>Just wondering--why didn't you just change the tests to use binary mode?
>
>
Ah, that's an easy one. I tried and failed.
It's not just the tests themselves; the tree comparison code would also
have to read the files in binary mode. When I did that, other tests
started failing because they new contents in text mode. Et cetera ad
nauseam. So instead of fixint the whole test suite to be binary-safe, I
threw away that lovely PNG.
Yes, binary-vs.-text is an interesting subject. I'm waiting for the day
when someone will start building Subvesion on Netware ... or OS/2 ... or
S/390 ... all of which are supported by APR. And S/390 is EBCDIC; no
less. Whee!
--
Brane Čibej <brane_at_xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Dec 5 23:58:53 2002