tech-userlevel archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Add curses_version() in curses(3)



On 02.09.2019 15:11, Valery Ushakov wrote:
> On Mon, Sep 02, 2019 at 12:32:51 +0300, Valery Ushakov wrote:
> 
>> Why would we ever want to report this completely random and unrelated
>> fact?!
>>
>> There were years when curses in the tree was unchanged.  In the mean
>> time we have churned through dozens of netbsd versions.
>>
>> Why are we trying to over-engineer this?
> 
> So the original claim was that it's needed by qemu.

For the reference. I noted that I originally needed it, not that I still
need it in qemu. I have patched upstream qemu to stop using it.

https://github.com/qemu/qemu/commit/271f37abb510607ca7650e40951284692a67579a

Although there are qemu forks in the wild; some of them still need it
(at least hqemu is interesting to me).

>  I've downloaded
> qemu-3.1.1, qemu-4.1.0, and qemu.git and I don't see it mentioned
> anywhere, grep -r curses_version qemu* returns nothing.
> 
> I've searched github and I see manual pages to curses_version in
> various formats, binding for curses_version for various scripting
> languages.  configure tests that seems to check the presense of
> curses_version to detect ncurses.
> 
> NB: have we just broken all those configure scripts?
> 
> I don't see this function actually being used for anything though I
> haven't clicked through all the 31K occurrences.
> 
> I cannot really conceive how this function can be useful for anything
> other that printing that information as part of some banner.  A
> program that makes some kind of decision about how to use curses by
> inspecting this value at runtime?  I'd sooner belive in unicorns :)
> 
> My preference would be to either revert this and pretend it never
> happened or to make it return a static string "All your base are
> belong to us" b/c we really don't have any meaningful versioning for
> our curses and pretending otherwise by returning completely unrelated
> netbsd version just makes us look stupid, IMO.  (As Christoph noted,
> do we now have to bump netbsd version if we make a change in curses?
> :)
> 
> I'm sorry I should have joined this bikeshed earlier, but as I said I
> didn't realize what was actually going to be committed.  I'm sorry I
> let Roy to be mobbed into this.
> 
> -uwe
> 

I'm for a static string like "NetBSD Curses"/"NetBSD-Curses" or even
"NetBSD" (comparable to "SVR4" in SVR4 Curses"), but Roy insisted on a
numerical version.

netbsd-curses a downstream fork uses its own versioning model and we do
not need to bother with any numbers locally.

The current approach is a consensus in that matter.

Attachment: signature.asc
Description: OpenPGP digital signature



Home | Main Index | Thread Index | Old Index