Hmm. It seems that, although RDA includes code for Cygwin and Solaris, that code is not actually built by default when one configures the tree in the normal way on those platforms. All you get is the librda library. This means that we can't know if regenerating the auto* files introduces additional build problems for that platform-specific code without first making RDA actually build it again. I don't want to extend the scope of my project to include making the Cygwin and Solaris native support code build again. But evolving the surrounding support will inevitably bit-rot that stuff. It's the classic "unmaintained code" dilemma. Ideally, that stuff were made to build again, but limiting ourselves to actions we can afford to take immediately, what should our policy be? Here are the options I see, listed in order of decreasing preference for me: a) Declare Cygwin and Solaris native support to be unmaintained in the README file, but leave the sources in the tree. b) Delete the Cygwin and Solaris native support. If someone wants to resurrect it, it's all in CVS. c) Put off upgrading the auto* files until Cygwin and Solaris native build again and the upgrade can be tested. How do other folks feel? In January, I'm going to have some time available to make a web page for RDA and put together a release. I don't know if there will be any public interest in RDA, but if there is, we might find volunteers to work on the Solaris and Cygwin ports.
On 08 Dec 2004 14:22:18 -0500
Jim Blandy <jimb@redhat.com> wrote:
> Hmm. It seems that, although RDA includes code for Cygwin and
> Solaris, that code is not actually built by default when one
> configures the tree in the normal way on those platforms. All you get
> is the librda library.
>
> This means that we can't know if regenerating the auto* files
> introduces additional build problems for that platform-specific code
> without first making RDA actually build it again. I don't want to
> extend the scope of my project to include making the Cygwin and
> Solaris native support code build again. But evolving the surrounding
> support will inevitably bit-rot that stuff. It's the classic
> "unmaintained code" dilemma.
>
> Ideally, that stuff were made to build again, but limiting ourselves
> to actions we can afford to take immediately, what should our policy
> be? Here are the options I see, listed in order of decreasing
> preference for me:
>
> a) Declare Cygwin and Solaris native support to be unmaintained in the
> README file, but leave the sources in the tree.
>
> b) Delete the Cygwin and Solaris native support. If someone wants to
> resurrect it, it's all in CVS.
>
> c) Put off upgrading the auto* files until Cygwin and Solaris native
> build again and the upgrade can be tested.
>
> How do other folks feel?
I vote for (a).
Kevin
On Wed, Dec 08, 2004 at 01:12:41PM -0700, Kevin Buettner wrote:
>On 08 Dec 2004 14:22:18 -0500
>Jim Blandy <jimb@redhat.com> wrote:
>
>> Hmm. It seems that, although RDA includes code for Cygwin and
>> Solaris, that code is not actually built by default when one
>> configures the tree in the normal way on those platforms. All you get
>> is the librda library.
>>
>> This means that we can't know if regenerating the auto* files
>> introduces additional build problems for that platform-specific code
>> without first making RDA actually build it again. I don't want to
>> extend the scope of my project to include making the Cygwin and
>> Solaris native support code build again. But evolving the surrounding
>> support will inevitably bit-rot that stuff. It's the classic
>> "unmaintained code" dilemma.
>>
>> Ideally, that stuff were made to build again, but limiting ourselves
>> to actions we can afford to take immediately, what should our policy
>> be? Here are the options I see, listed in order of decreasing
>> preference for me:
>>
>> a) Declare Cygwin and Solaris native support to be unmaintained in the
>> README file, but leave the sources in the tree.
>>
>> b) Delete the Cygwin and Solaris native support. If someone wants to
>> resurrect it, it's all in CVS.
>>
>> c) Put off upgrading the auto* files until Cygwin and Solaris native
>> build again and the upgrade can be tested.
>>
>> How do other folks feel?
>
>I vote for (a).
I thought Corinna Vinschen was maintaining rda for cygwin.
cgf
Christopher Faylor <me@cgf.cx> writes:
> On Wed, Dec 08, 2004 at 01:12:41PM -0700, Kevin Buettner wrote:
> >On 08 Dec 2004 14:22:18 -0500
> >Jim Blandy <jimb@redhat.com> wrote:
> >
> >> Hmm. It seems that, although RDA includes code for Cygwin and
> >> Solaris, that code is not actually built by default when one
> >> configures the tree in the normal way on those platforms. All you get
> >> is the librda library.
> >>
> >> This means that we can't know if regenerating the auto* files
> >> introduces additional build problems for that platform-specific code
> >> without first making RDA actually build it again. I don't want to
> >> extend the scope of my project to include making the Cygwin and
> >> Solaris native support code build again. But evolving the surrounding
> >> support will inevitably bit-rot that stuff. It's the classic
> >> "unmaintained code" dilemma.
> >>
> >> Ideally, that stuff were made to build again, but limiting ourselves
> >> to actions we can afford to take immediately, what should our policy
> >> be? Here are the options I see, listed in order of decreasing
> >> preference for me:
> >>
> >> a) Declare Cygwin and Solaris native support to be unmaintained in the
> >> README file, but leave the sources in the tree.
> >>
> >> b) Delete the Cygwin and Solaris native support. If someone wants to
> >> resurrect it, it's all in CVS.
> >>
> >> c) Put off upgrading the auto* files until Cygwin and Solaris native
> >> build again and the upgrade can be tested.
> >>
> >> How do other folks feel?
> >
> >I vote for (a).
>
> I thought Corinna Vinschen was maintaining rda for cygwin.
Oh! I'll ask her about it, then. She's been on vacation, which may
be why she hasn't spoken up.