On Wed, 7 Apr 2021, Liu Hao wrote:
在 2021/4/6 下午10:15, Martin Storsjö 写道:
Without having read the patch that is being discussed... While that header
does define them, for new code I generally prefer to use the GCC style
defines, for consistency (or in pairs with the MSVC style one), and to
在 2021/4/6 下午10:15, Martin Storsjö 写道:
Without having read the patch that is being discussed... While that header does define them, for new
code I generally prefer to use the GCC style defines, for consistency (or in pairs with the MSVC
style one), and to reduce the risk of accidentally using
在 2021/4/6 下午10:04, Martin Storsjö 写道:
(I guess it's not entirely unconditional - the user can still set the _USE_32BIT_TIME_T - but that's
a technicality.)
Yeah I was thinking about phrases such as 'both default' but mistook 'unconditional'... Anyway this
word doesn't appear in the commit
On Tue, 6 Apr 2021, Liu Hao wrote:
在 2021-04-01 22:33, Liu Hao 写道:
在 2021-04-01 22:09, Biswapriyo Nath 写道:
* Another query, why does this header use _M_IX86 like
manifest
constants? I thought those are only for MSVC and we uses
__i686__
instead.
I think those should be changed to GCC one
On Tue, 6 Apr 2021, Liu Hao wrote:
在 2021-03-31 21:45, Liu Hao 写道:
Side note: This patch is not meant to address the issue
about default size of `time_t`, due to possibility of ABI
break.
And here is the second patch that makes `time_t`
unconditionally 64-bit for UCRT.
(I guess it's not
在 2021-04-01 22:33, Liu Hao 写道:
在 2021-04-01 22:09, Biswapriyo Nath 写道:
* Another query, why does this header use _M_IX86 like manifest
constants? I thought those are only for MSVC and we uses __i686__
instead.
I think those should be changed to GCC ones, too.
Just checked. Those MSVC-is
在 2021-03-31 21:45, Liu Hao 写道:
Side note: This patch is not meant to address the issue about default size of `time_t`, due to
possibility of ABI break.
And here is the second patch that makes `time_t` unconditionally 64-bit for
UCRT.
--
Best regards,
Liu Hao
From 00e93ac9d535f4f7e285e706b
在 2021-04-06 04:22, Martin Storsjö 写道:
The patch looks good to me, thanks!
Should we change the default target for unsuffixed aliases in import libraries then, too, even if
they're primarily used for configure checks?
I'll send a new patch that changes both the header and DEFs.
--
Best r
