Re: [PATCH 1/3] t3404: preserve test_tick state across short SHA-1 collision test

2013-08-25 Thread Eric Sunshine
On Sun, Aug 25, 2013 at 1:55 AM, Jonathan Nieder jrnie...@gmail.com wrote: Hi, Eric Sunshine wrote: The short SHA-1 collision test requires carefully crafted commits in order to ensure a collision at rebase time. Yeah, this breaks the usual rule that tests should be independent of hashing

Re: [PATCH 1/3] t3404: preserve test_tick state across short SHA-1 collision test

2013-08-25 Thread Jonathan Nieder
Eric Sunshine wrote: On Sun, Aug 25, 2013 at 1:55 AM, Jonathan Nieder jrnie...@gmail.com wrote: Would be clearer if the code in a subshell were indented: ( unset test_tick test_commit ... ) I considered it, but decided against it for a

Re: [PATCH 1/3] t3404: preserve test_tick state across short SHA-1 collision test

2013-08-25 Thread Junio C Hamano
Jonathan Nieder jrnie...@gmail.com writes: Eric Sunshine wrote: On Sun, Aug 25, 2013 at 1:55 AM, Jonathan Nieder jrnie...@gmail.com wrote: Would be clearer if the code in a subshell were indented: ( unset test_tick test_commit ... ) I

Re: [PATCH 1/3] t3404: preserve test_tick state across short SHA-1 collision test

2013-08-24 Thread Jonathan Nieder
Hi, Eric Sunshine wrote: The short SHA-1 collision test requires carefully crafted commits in order to ensure a collision at rebase time. Yeah, this breaks the usual rule that tests should be independent of hashing function. But it's the best we can do, I think. [...] ---

[PATCH 1/3] t3404: preserve test_tick state across short SHA-1 collision test

2013-08-21 Thread Eric Sunshine
The short SHA-1 collision test requires carefully crafted commits in order to ensure a collision at rebase time. This involves managing state which impacts the resulting SHA-1, including commit time. To accomplish this, test_tick is set to a known state for the short SHA-1 collision test.