Re: [bug] mercurial 0.9.5
On Fri, Dec 28, 2007 at 03:09:09PM +0100, Marc Espie wrote:
> Strongly disagree. This stuff makes all the difference in the
> world. As far as possible, stuff should work out of the box on
> OpenBSD.
OK -- I'm swayed. I have a few partial patches hanging around that
I'm integrating now/next week. I post something back here soonish.
Thanks!
--
o--{ Will Maier }--o
| web:...http://www.lfod.us/ | [EMAIL PROTECTED] |
*--[ BSD Unix: Live Free or Die ]--*
Re: [bug] mercurial 0.9.5
Marc Espie wrote: Strongly disagree. This stuff makes all the difference in the world. As far as possible, stuff should work out of the box on OpenBSD. I spend a lot of time with a lot of software. One strong appeal of OpenBSD is that stuff just works. As opposed to linux, where you almost always have to go a mailing-list, or fiddle with your configuration to make things work. When you're young enough, you derive a sense of accomplishment from configuring things just right. When you become an old fart, you just have interesting work to be done, and stupid configuration details just get in the way (one thing which is hard to grasp for computer geeks is that some users just want to do their work, and don't care at all about configuration details). ... You could extend the OpenBSD motto `secure by default' to `works by default'. I agree with Marc, things should work as much as possible. The original issue was wish8.4 (installed) vs "wish" (expected). We have a similar situation with python, where it installs in /usr/local/bin/ python2.4 or python2.5 and asks you to symlink python yourself. Versioning is a general problem. OpenBSD solved something similar with /etc/mailer.conf and the binary /usr/sbin/mailwrapper, but only for different sendmail programs, not for differently-numbered versions of the same program. I wonder if this could be generalized somewhat, to produce a functionality similar to (but simpler) than Linux' "alternatives" mechanism? How much simpler? The ports MODULES already know about different versions of the programs so they could conceivably auto-update the symlink to the highest installed version? I know this won't work for 100% of the cases, but if it "just works" for most of them . . .
Re: [bug] mercurial 0.9.5
Strongly disagree. This stuff makes all the difference in the world. As far as possible, stuff should work out of the box on OpenBSD. I spend a lot of time with a lot of software. One strong appeal of OpenBSD is that stuff just works. As opposed to linux, where you almost always have to go a mailing-list, or fiddle with your configuration to make things work. When you're young enough, you derive a sense of accomplishment from configuring things just right. When you become an old fart, you just have interesting work to be done, and stupid configuration details just get in the way (one thing which is hard to grasp for computer geeks is that some users just want to do their work, and don't care at all about configuration details). The likelyhood of bitrot is not really big, those changes are reasonably easy to maintain, on one hand. On the other hand, maintainers should communicate with upstream developers so that this situation is resolved correctly. You could extend the OpenBSD motto `secure by default' to `works by default'. It's the same philosophy. Really.
Re: [bug] mercurial 0.9.5
On Dec 27, 2007 5:19 PM, Will Maier <[EMAIL PROTECTED]> wrote:
> On Thu, Dec 27, 2007 at 04:55:17PM +0100, Benoit Chesneau wrote:
> > little bug in current mercurial package. hgk extension use wish
> > (installed with tk) but wish8.4 is installed on openbsd. Here is a
> > little patch to fix it.
>
> I like the intent of this patch, but I've considered and decided
> against similar things in the past. A patch like this would provide
> only a little benefit (since users already have to install tk and
> move ${PREFIX}/share/mercurial/hgk into their $PATH anyway) while
> increasing the likelihood of bitrot in devel/mercurial.
>
> As things stand, the port installs several files under ${PREFIX}/share
> that may require some fiddling before they'll work. I think this is
> a reasonable compromise between providing extra functionality (like
> zsh_completion or mercurial.el) without making the mercurial package
> a PITA to install.
>
> That said, I'm still open to other opinions on this, though.
>
Well you don't need to move any file to use hgk. Customizing ~/.hgrc like this :
--
[extensions]
hgk=/usr/local/lib/python2.5/site-packages/hgext/hgk.py
[hgk]
path=/usr/local/share/mercurial/hgk
is enough. But think you're right, this is really little benefit. I
hesitated before
sending the patch.
- benoît
Re: [bug] mercurial 0.9.5
On Thu, Dec 27, 2007 at 04:55:17PM +0100, Benoit Chesneau wrote:
> little bug in current mercurial package. hgk extension use wish
> (installed with tk) but wish8.4 is installed on openbsd. Here is a
> little patch to fix it.
I like the intent of this patch, but I've considered and decided
against similar things in the past. A patch like this would provide
only a little benefit (since users already have to install tk and
move ${PREFIX}/share/mercurial/hgk into their $PATH anyway) while
increasing the likelihood of bitrot in devel/mercurial.
As things stand, the port installs several files under ${PREFIX}/share
that may require some fiddling before they'll work. I think this is
a reasonable compromise between providing extra functionality (like
zsh_completion or mercurial.el) without making the mercurial package
a PITA to install.
That said, I'm still open to other opinions on this, though.
--
o--{ Will Maier }--o
| web:...http://www.lfod.us/ | [EMAIL PROTECTED] |
*--[ BSD Unix: Live Free or Die ]--*
[bug] mercurial 0.9.5
little bug in current mercurial package. hgk extension use wish (installed with tk) but wish8.4 is installed on openbsd. Here is a little patch to fix it. - benoît -- http://friendpaste.com mercurial.diff Description: Binary data
