On Tue 16th Sep, 2008 at 21:08, Bram Moolenaar seems to have written:
> Clea F. Rees wrote:
>>>> Definitely something in my environment. I rebuilt 7.314 and make test
>>>> produced the same failures. Then I took a copy of the source tree I
>>>> produced when I *originally* compiled 7.314, ran just "make testclean"
>>>> followed by "make test" and got the same errors - even though the very
>>>> same binaries originally passed all the tests.
>>>> I'm just wondering if anybody can suggest where I should start looking.
>>>> Can somebody tell me what tests 58 and 59 test? I don't understand the
>>>> code very well. Is there something special/different about 58 and 59?
>>>> Do they maybe have something in common which the others don't that
>>>> might give me a clue?
>>> You could check if the input files didn't get corrupted.
>>> The tests are for spell file features. This particular part is for
>>> sound folding. But since one of the three lines work it appears the
>>> soundfolding is working.
>> Hmmm... I'll look it up later.
>> For now: definitely not file corruption. It has something to do with
>> running the test suite in GNU screen. No matter which options I compile
>> with - and even if I go back to the binaries I compiled before for
>> 7.314 which passed all tests at the time - I get the same failures.
>> *Except* if I run the tests outside of GNU screen.
>> Yet I'm pretty sure I would have run under screen before - not 100%
>> sure, though. I don't know enough to pin it down though...
> That is really weird. I can't see why using screen would break only
> this part of this test.
I agree about the weirdness. I'm not even sure what made me try -
except I started trying a lot of things after I discovered binaries I
compiled weeks or months ago were failing in the same way when they'd
passed originally. I think I quit screen as the first step in an
"eliminate all custom environment settings" strategy. But simply
quitting screen turned out to be enough...
Is there any helpful information I might send anybody which might throw
any light on this? Or anything I might do to investigate?
You received this message from the "vim_mac" maillist.
For more information, visit http://www.vim.org/maillist.php