[arch-general] Haskell Support Was: [arch-haskell] Xmonad version?
Haskell platform was released with the latest stable GHC (7.0.4 of course). Guys, what's happening with haskell support in Arch? Xmonad, and now GHC and lots of packages in the supported repos are not keeping update with upstream.
On 12/17/2011 11:41 PM, Bernardo Barros wrote:
Haskell platform was released with the latest stable GHC (7.0.4 of course).
Guys, what's happening with haskell support in Arch?
Xmonad, and now GHC and lots of packages in the supported repos are not keeping update with upstream.
i vote for dropping xmonad and all packages to aur and let the community handle them. it seems we are not doing a great job at keeping them up to date. -- Ionuț
There was a recent Arch-Haskell thread about dumping most of AUR Haskell packages: http://www.haskell.org/pipermail/arch-haskell/2011-November/001744.html As an end user, I effectively ignore the official Arch Haskell packages as they are so far out of date. I'm also now trying to disregard most of the AUR since I came across that thread. Haskell is starting to feel fragile on Arch. What about moving *all* Haskell related packages to the Haskell Arch repo? All the Haskell Platform stuff, alex, happy, etc. Ethan Schoonover Github/Freenode: altercation - Solarized: http://ethanschoonover.com/solarized On Sat, Dec 17, 2011 at 13:44, Ionut Biru <biru.ionut@gmail.com> wrote:
On 12/17/2011 11:41 PM, Bernardo Barros wrote:
Haskell platform was released with the latest stable GHC (7.0.4 of course).
Guys, what's happening with haskell support in Arch?
Xmonad, and now GHC and lots of packages in the supported repos are not keeping update with upstream.
i vote for dropping xmonad and all packages to aur and let the community handle them.
it seems we are not doing a great job at keeping them up to date.
-- Ionuț
On 17/12/11 23:07, Ethan Schoonover wrote:
There was a recent Arch-Haskell thread about dumping most of AUR Haskell packages: http://www.haskell.org/pipermail/arch-haskell/2011-November/001744.html
As an end user, I effectively ignore the official Arch Haskell packages as they are so far out of date. I'm also now trying to disregard most of the AUR since I came across that thread. Haskell is starting to feel fragile on Arch.
What about moving *all* Haskell related packages to the Haskell Arch repo? All the Haskell Platform stuff, alex, happy, etc.
Ethan Schoonover Github/Freenode: altercation - Solarized: http://ethanschoonover.com/solarized
On Sat, Dec 17, 2011 at 13:44, Ionut Biru <biru.ionut@gmail.com> wrote:
On 12/17/2011 11:41 PM, Bernardo Barros wrote:
Haskell platform was released with the latest stable GHC (7.0.4 of course).
Guys, what's happening with haskell support in Arch?
Xmonad, and now GHC and lots of packages in the supported repos are not keeping update with upstream.
i vote for dropping xmonad and all packages to aur and let the community handle them.
it seems we are not doing a great job at keeping them up to date.
-- Ionuț Hi,
What are the real troubling haskell packages? Only GHC, haskell-platform and XMonad? I'll speak with Vesa about XMonad, i might be able to adopt it. -- Jelle van der Waa
On 18/12/11 09:53, Jelle van der Waa wrote:
On 17/12/11 23:07, Ethan Schoonover wrote:
There was a recent Arch-Haskell thread about dumping most of AUR Haskell packages: http://www.haskell.org/pipermail/arch-haskell/2011-November/001744.html
As an end user, I effectively ignore the official Arch Haskell packages as they are so far out of date. I'm also now trying to disregard most of the AUR since I came across that thread. Haskell is starting to feel fragile on Arch.
What about moving *all* Haskell related packages to the Haskell Arch repo? All the Haskell Platform stuff, alex, happy, etc.
Ethan Schoonover Github/Freenode: altercation - Solarized: http://ethanschoonover.com/solarized
On Sat, Dec 17, 2011 at 13:44, Ionut Biru <biru.ionut@gmail.com> wrote:
On 12/17/2011 11:41 PM, Bernardo Barros wrote:
Haskell platform was released with the latest stable GHC (7.0.4 of course).
Guys, what's happening with haskell support in Arch?
Xmonad, and now GHC and lots of packages in the supported repos are not keeping update with upstream.
i vote for dropping xmonad and all packages to aur and let the community handle them.
it seems we are not doing a great job at keeping them up to date.
-- Ionuț Hi,
What are the real troubling haskell packages? Only GHC, haskell-platform and XMonad?
I'll speak with Vesa about XMonad, i might be able to adopt it.
Btw adopted xmonad,xmonad-contrib, these packages will be updated today. -- Jelle van der Waa
On Sun, Dec 18, 2011 at 11:31:09AM +0100, Jelle van der Waa wrote: [...]
Btw adopted xmonad,xmonad-contrib, these packages will be updated today.
Excellent, then I can stop building these for myself. Are you considering adding xmonad-extras too, or should I go ahead and add that to [haskell] instead? /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus Perl is another example of filling a tiny, short-term need, and then being a real problem in the longer term. -- Alan Kay
On 18/12/11 11:48, Magnus Therning wrote:
On Sun, Dec 18, 2011 at 11:31:09AM +0100, Jelle van der Waa wrote: [...]
Btw adopted xmonad,xmonad-contrib, these packages will be updated today.
Excellent, then I can stop building these for myself. Are you considering adding xmonad-extras too, or should I go ahead and add that to [haskell] instead?
/M
Hmm didn't follow the mailing list about xmonad-extras. If it's easy to build and doesn't need real strange dependencies I could add it. Does it bring nice new features ? :) -- Jelle van der Waa
On Sun, Dec 18, 2011 at 11:53:29AM +0100, Jelle van der Waa wrote:
On 18/12/11 11:48, Magnus Therning wrote:
On Sun, Dec 18, 2011 at 11:31:09AM +0100, Jelle van der Waa wrote: [...]
Btw adopted xmonad,xmonad-contrib, these packages will be updated today.
Excellent, then I can stop building these for myself. Are you considering adding xmonad-extras too, or should I go ahead and add that to [haskell] instead?
/M
Hmm didn't follow the mailing list about xmonad-extras. If it's easy to build and doesn't need real strange dependencies I could add it. Does it bring nice new features ? :)
It's straight forward to build. It required no modules that weren't already in [haskell] (I haven't checked that for [community] though). ATM there's only one thing I'm using from it in my setup: XMonad.Actions.Volume: http://hackage.haskell.org/packages/archive/xmonad-extras/0.10.1/doc/html/XM... Adding it to [haskell] takes less than a minute so I don't mind adding it there :-) /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus I invented the term Object-Oriented, and I can tell you I did not have C++ in mind. -- Alan Kay
Magnus Therning [2011.12.18 1148 +0100]:
On Sun, Dec 18, 2011 at 11:31:09AM +0100, Jelle van der Waa wrote: [...]
Btw adopted xmonad,xmonad-contrib, these packages will be updated today.
Excellent, then I can stop building these for myself. Are you considering adding xmonad-extras too, or should I go ahead and add that to [haskell] instead?
I'm not sure whether you really want to stop building xmonad for yourself. With new management upstream, things may change from here on, but in the past there were very long intervals between official releases, so that most people ended up running the latest development version from darcs. The good thing is that, in spite of it being the development version, things hardly ever (in my case, not at all over the last 2 years) break. Cheers, Norbert
Just try to support Haskell Platform in extra/community. That's enough.
On Sun, Dec 18, 2011 at 11:31:09AM +0100, Jelle van der Waa wrote: [...]
Btw adopted xmonad,xmonad-contrib, these packages will be updated today.
When building xmonad-extras against these updated packages I noticed that you didn't include a version compiled with profiling enabled (maybe not such an important thing to have though), nor did you include any haddock documentation (that might be a useful thing to include). /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus I invented the term Object-Oriented, and I can tell you I did not have C++ in mind. -- Alan Kay
So is the current best practice for Haskell on Arch something like this: 0. Always prefer to source Haskell packages from official Arch repos now that they are updated 1. Always add the [haskell] repo to pacman.conf 2. Always avoid AUR Haskell packages (they are mostly out of date) 3. Always avoid plain cabal, use cabal2arch instead 4. Only if a package isn't in official repos or [haskell], use cabal2arch to create a package I know that this may seem like oversimplification, but despite following the various Haskell issues and lists closely I occasionally come away confused about current recommendations, and I'm not sure that the wiki is up to date on this issue. Thanks for any guidance, es Ethan Schoonover Github/Freenode: altercation - Solarized: http://ethanschoonover.com/solarized On Mon, Dec 19, 2011 at 12:59, Magnus Therning <magnus@therning.org> wrote:
On Sun, Dec 18, 2011 at 11:31:09AM +0100, Jelle van der Waa wrote: [...]
Btw adopted xmonad,xmonad-contrib, these packages will be updated today.
When building xmonad-extras against these updated packages I noticed that you didn't include a version compiled with profiling enabled (maybe not such an important thing to have though), nor did you include any haddock documentation (that might be a useful thing to include).
/M
-- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus
I invented the term Object-Oriented, and I can tell you I did not have C++ in mind. -- Alan Kay
On Tue, Dec 20, 2011 at 4:11 AM, Ethan Schoonover <es@ethanschoonover.com>wrote:
2. Always avoid AUR Haskell packages (they are mostly out of date)
I'm not sure about this. First I find this unrespectful to the people spending time maintaining Haskell PKGBUILDs in the AUR (and I'm not saying this before I do). Second, my own experience lead me to disable [haskell] repository on my computers because some packages were out of date but where picked before AUR ones by the aur helper I use. -- Cédric Girard
2011/12/20 Cédric Girard <girard.cedric@gmail.com>
On Tue, Dec 20, 2011 at 4:11 AM, Ethan Schoonover <es@ethanschoonover.com
wrote:
2. Always avoid AUR Haskell packages (they are mostly out of date)
I'm not sure about this. First I find this unrespectful to the people spending time maintaining Haskell PKGBUILDs in the AUR (and I'm not saying this before I do).
No disrespect intended; this comes directly from the arch-haskell mailing list: http://www.haskell.org/pipermail/arch-haskell/2011-November/001744.html Their point is specifically that many AUR haskell packages are not at all maintained. I'm sure there are some that are, but until this cull is complete or a "whitelist" of AUR haskell packages shows up, I don't know which AUR haskell packages those would be.
Second, my own experience lead me to disable [haskell] repository on my computers because some packages were out of date but where picked before AUR ones by the aur helper I use.
Hence the confusion around this issue. If Haskell AUR packages are indeed being deprecated then is [haskell] now more of a definitive source or is it *also* not well maintained? Take a look at that thread I link to in this mail, hopefully it will clarify my inclusion of the avoid AUR position in the rough package selection heuristic. es
On Tue, Dec 20, 2011 at 12:11:40PM -0800, Ethan Schoonover wrote:
2011/12/20 Cédric Girard <girard.cedric@gmail.com>
On Tue, Dec 20, 2011 at 4:11 AM, Ethan Schoonover <es@ethanschoonover.com> wrote:
2. Always avoid AUR Haskell packages (they are mostly out of date)
I'm not sure about this. First I find this unrespectful to the people spending time maintaining Haskell PKGBUILDs in the AUR (and I'm not saying this before I do).
No disrespect intended; this comes directly from the arch-haskell mailing list: http://www.haskell.org/pipermail/arch-haskell/2011-November/001744.html
Their point is specifically that many AUR haskell packages are not at all maintained. I'm sure there are some that are, but until this cull is complete or a "whitelist" of AUR haskell packages shows up, I don't know which AUR haskell packages those would be.
The cull is complete. I've orphaned *all* haskell packages on AUR that were owned by the archhaskell user.
Second, my own experience lead me to disable [haskell] repository on my computers because some packages were out of date but where picked before AUR ones by the aur helper I use.
Hence the confusion around this issue. If Haskell AUR packages are indeed being deprecated then is [haskell] now more of a definitive source or is it *also* not well maintained?
I'd say it's fairly well maintained. There are always going to be issues with tracking Hackage due to 1. The frequency of updates being uploaded to Hackage. 2. The dependencies of packages. 3. The decision to stick to HP in Arch [extra]/[community]. The amount of recompilation necessary in [haskell] is a bit daunting, but unfortunately necessary. This means that if you can offer access to a *fast* build machine then that'd be very useful. Even more useful would an automated build machine be; some place where source packages could be uploaded to be built.
Take a look at that thread I link to in this mail, hopefully it will clarify my inclusion of the avoid AUR position in the rough package selection heuristic.
/M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus Perl is another example of filling a tiny, short-term need, and then being a real problem in the longer term. -- Alan Kay
On Tue, Dec 20, 2011 at 9:48 PM, Magnus Therning wrote:
Second, my own experience lead me to disable [haskell] repository on my computers because some packages were out of date but where picked before AUR ones by the aur helper I use.
Hence the confusion around this issue. If Haskell AUR packages are indeed being deprecated then is [haskell] now more of a definitive source or is it *also* not well maintained?
I'd say it's fairly well maintained.
Well maybe my experience on this is not the latest. Last time I tried [haskell] was some months ago and I believe there was some changes in the arch-haskell team since. Anyway some of the old arch-haskell PKGBUILDs in the AUR have been adopted since the mass orphan and I think the situation will get better there as maintenance will be spread among more people. -- Cédric Girard
On Wed, Dec 21, 2011 at 12:00:12AM +0100, Cédric Girard wrote:
On Tue, Dec 20, 2011 at 9:48 PM, Magnus Therning wrote:
Second, my own experience lead me to disable [haskell] repository on my computers because some packages were out of date but where picked before AUR ones by the aur helper I use.
Hence the confusion around this issue. If Haskell AUR packages are indeed being deprecated then is [haskell] now more of a definitive source or is it *also* not well maintained?
I'd say it's fairly well maintained.
Well maybe my experience on this is not the latest. Last time I tried [haskell] was some months ago and I believe there was some changes in the arch-haskell team since.
Anyway some of the old arch-haskell PKGBUILDs in the AUR have been adopted since the mass orphan and I think the situation will get better there as maintenance will be spread among more people.
I on the other hand suspect it won't get much better at all :-) There is a real risk that having several maintainers will result in packages that, while they are up-to-date, don't play well together. /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus Most software today is very much like an Egyptian pyramid with millions of bricks piled on top of each other, with no structural integrity, but just done by brute force and thousands of slaves. -- Alan Kay
On Sun, Dec 18, 2011 at 09:53:40AM +0100, Jelle van der Waa wrote: [...]
What are the real troubling haskell packages? Only GHC, haskell-platform and XMonad?
I'm not sure any of them are troubling. This is my view on it: - GHC is a big package, but Vesa is doing a good job keeping it up-to-date. Currently it *looks* like it lagging behind upstream, but that's caused by a strange move to use an even/release version number for a tech-preview release. - IMNSHO HP doesn't have a place in Arch, it should be dropped and instead the individual packages should be updated as new versions are available. See a recent discussion on the ArchHaskell mailing list. - Xmonad is far from troubling, it just needs an update that for some reason have been overlooked. /M -- Magnus Therning OpenPGP: 0xAB4DFBA4 email: magnus@therning.org jabber: magnus@therning.org twitter: magthe http://therning.org/magnus Perl is another example of filling a tiny, short-term need, and then being a real problem in the longer term. -- Alan Kay
participants (7)
-
Bernardo Barros
-
Cédric Girard
-
Ethan Schoonover
-
Ionut Biru
-
Jelle van der Waa
-
Magnus Therning
-
Norbert Zeh