apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <wr...@rowe-clan.net>
Subject Re: HEAD and releases
Date Wed, 16 Apr 2003 07:56:43 GMT
At 02:11 AM 4/16/2003, Greg Stein wrote:
>On Wed, Apr 16, 2003 at 01:06:44AM +0100, Joe Orton wrote:
>>...
>> I don't understand how APR will ever become 1.0-worthy if HEAD remains
>> indefinitely constrained to being 0.9.2-compatible.
>
>I agree. We will never get to 1.0 if we keep worrying about compatibility so
>much. If we call it 1.0 then we have a legitimate need to continue
>compatibility. But right now, we're stuck in limbo.

And we will have proven we are utterly incapable of doing so.  Hoorah.

I am waiting for an example of something that must be broken, as opposed
to marked /* @deprecated */ for DoxyGen.

An example of your limbo, please? Enlighten me, or split APR_0_9_BRANCH 
already let's start spreading all of the bug fixes across two trees.  I don't
think that's a bad idea...

if we believe most of the bugs have been squished, and we are *happy*
with the API for the next 18 months or so, then it's time already for 1.0

a quick browse through STATUS convinces me we haven't, but that's
an opinion, not any sort of vote.

Would anyone who is asking to break binary compat paus just long enough
to explain one bit of the API that requires the sort of surgery that breaks 
binary compat?  I have one example; in the transparent apr_finfo_t structure,
we need to distinguish ctime from 'crtime' - or file creation time on those
platforms that support it.  Is it worth breaking compat?  Not really.

Any others?  I'm all ears.

Bill



Mime
View raw message