DMD-0.151 regressions

[Thomas Kuehne]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Stewart Gordon schrieb am 2006-04-06:
> Thomas Kuehne wrote:
>> Stewart Gordon schrieb am 2006-04-06:

<snip>

> I see.  But what does "position independent code" mean?

from GCC's man page:
> -fPIC
>	If supported for the target machine, emit position-independent
>	code, suitable for dynamic linking and avoiding any limit on the
>	size of the global offset table. 

<snip>

> Looking at 0.138, the lowest number of passes is 1299, but you've taken 
> 1304 as the figure for that version.  Time to investigate the formula....

You are looking at revision 1108 (2006-03-21) instead of the
current 1133 (2006-04-05) one.

The formula looks for the worst test result of each test case.
Naturaly "untested" isn't considered a bad result.
There are 44 untested cases in the "-g -fPIC -O" configuration,
thus the highest possible number of global passes could have been
1343 (1299+44).

1343 >= 1304 >= 1299: sanity check passed

Thomas


-----BEGIN PGP SIGNATURE-----

iD4DBQFENY6L3w+/yD4P9tIRAtj8AKCxnkwyfeLwvDG5I7/NyVPhlkUVxACY7kom
QZy15zaSRYo1A4S4zof38Q==
=ERet
-----END PGP SIGNATURE-----