On 11/05/2011 07:32 PM, Brandon Allbery wrote:

You're running stock 0.9.1/0.9.2, which is effectively 2 years old at this point. Upgrade to the darcs version;

..? Really? Can't you just do a quarterly drop or something? Or even annual? - Allen S. ROut

Tom Davis [2011.11.07 1122 -0500]:

On Nov 7, 2011, at 10:29 AM, "Allen S. Rout" wrote:

...

On 11/05/2011 07:32 PM, Brandon Allbery wrote:

...

You're running stock 0.9.1/0.9.2, which is effectively 2 years old at this point. Upgrade to the darcs version;

..? Really? Can't you just do a quarterly drop or something? Or even annual?

I definitely agree. I had no idea until recently that the Fedora package the xmonad site linked me to was effectively obsolete. Or, rather, that darcs was under active development since. Now I'm not sure what will happen if I upgrade to darcs, if it's worth it, etc.

I guess I don't consider "bleeding edge" to be synonymous with "the past two years of development." If darcs is canonical and stable at this point, it should be presented as the preferred installation method at the very least.

Let me address the two separate points here: (1) No official release in 2 years: This makes me a bit uneasy about the future of xmonad, as it seems the original authors have lost interest in the project. In only hope, and there is some evidence for that, that the community that has built up around xmonad includes people who are willing to ensure the survival of this excellent piece of software. In any case, Gwern Branwen keeps nagging on the list that V1.0 should finally be released. I hope he succeeds. (2) Running the darcs version: Wanting to contribute code here and there shortly after switching to xmonad, I was advised to run the darcs version as opposed to the then official release, in order not to duplicate work done by others or introduce incompatible patches. I was concerned about stability (in terms of API and in terms of no crashes), but I was assured that both are very stable even in the darcs version - the only things that change are under-the-hood bugfixes and addition of new modules. I have learned over the years that this assessment was entirely correct - solid and stable as a rock. In particular, if you're worried about your configuration file being hosed as a result of pulling in the latest changes - that's almost guaranteed not to happen because the xmonad community is in general extremely careful not to introduce API changes that break existing setups. Bottom line: I highly recommend running the darcs version, particularly given the current state of affairs regarding point (1). I hope this helps. Cheers, Norbert

The *original* xmonad developers have moved on. The current crop is quite active... but doesn't have access bits on the archive. We're still working on that (and in fact someone who was given access some time back and promptly got sandbagged by Real Life just swapped back in, so things may finally be about to move). How about a release every 6 months? Someone mentioned this already and I

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Brandon, there are more issues here and I got an opinion about xmonads release politics too.. but I'll keep that on topic. On 07.11.2011 18:22, Brandon Allbery wrote: think it makes sense. New users would see that the project is under active development. In addition they would be able to get some very recent code without having to fiddle with darcs etc. I'm running xmonad since around 2008 and I think it would be a shame if it would go down the drain because it cannot draw new users (and probably developers). XMonad made me learn Haskell and FP in general and I'm very grateful for that. If there's some work to be done, I'd be glad to help out.

(As to the original question, I still haven't had a chance to check what's going on but I had thought I'd verified at one point that WM_S0 was being created normally. Then again, I was running a rather strange debugging setup at the time.)

I tried the xmonad version from darcs I mentioned in my last email. Using X11.Graphics.Xlib methods [1] as well as 'xlsatoms' [2] reveals that there is no WM_S* atom. So, as I said before, xmonad does not set the atom properly on initial startup (e.g. by using 'exec xmonad' in .xinitrc). Best regards, Jochen [1] atom :: IO Atom atom = openDisplay ":0.0" >>= \dpy -> internAtom dpy ("WM_S" ++ show (defaultScreen dpy)) True main = atom >>= print [2] [852 ~$] DISPLAY=:0.0 xlsatoms | grep -E 'WM_S?$' [853 ~$] -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iEYEARECAAYFAk64VAcACgkQtVwvsA+W4CAINACghxZzsugjugtXcUIquEz1ngmd yEUAn2d5Yl3XkohFfBY732CfLnJNySTk =Zuvq -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, On 08.11.2011 00:04, Ivan Lazar Miljenovic wrote:

On 8 November 2011 08:56, Jochen Keil wrote:

...

How about a release every 6 months? Someone mentioned this already and I think it makes sense. New users would see that the project is under active development. In addition they would be able to get some very recent code without having to fiddle with darcs etc.

I'm not a fan of "let's release just because we need a release every 6 months": I think a feature-based release ("do we have enough changes to warrant a release?") makes more sense, albeit with the uncertainty of a looming deadline to make people consider releasing.

I agree with your point about making a release without changes. However, there were enough changes, patches and bugfixes in the last two years to justify a release. This is also some kind of a chicken-egg problem: Without new releases no new users and developers get attracted hence no changes ultimately leading to stagnation. In a project like xmonad there will be enough changes during six months to have a new release. And if it's only to check if it still compiles and to update the tarballs. It's more important to show that the project is still alive. Personally I'd also vote for release versions like 2011Q1 or 20110301 or something similar (just propagate if it's yyyymmdd or yyyyddmm ;). In case of urgent bugfixes you could still do something like 2011Q1.01 or 20110301.01 etc. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) iEYEARECAAYFAk646hkACgkQtVwvsA+W4CBrMwCfaE3ITdMXvH/aqzdQw0xHnkVj MkEAnjZm9/XOX4nmjUx8MeW1q6q1vpnH =e8ky -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 08.11.2011 09:48, Ivan Lazar Miljenovic wrote:

On 8 November 2011 19:36, Jochen Keil wrote:

...

I agree with your point about making a release without changes. However, there were enough changes, patches and bugfixes in the last two years to justify a release. This is also some kind of a chicken-egg problem:

Oh, it's definitely worth having one _now_. Sure, but why isn't it happening? :) What's keeping everyone from releasing the long debated (there were some mails a few weeks ago..) 1.0 version? I mean it can't be too hard to tag 1.0 in darcs and build a tarball. I'd do it myself if a) I'd be a bit more involved in xmonad development and b) this wouldn't feel like just pushing my name to the front now.

...

Personally I'd also vote for release versions like 2011Q1 or 20110301 or something similar (just propagate if it's yyyymmdd or yyyyddmm ;). In case of urgent bugfixes you could still do something like 2011Q1.01 or 20110301.01 etc.

Except we can't use these types of version "numbers" for Hackage, and they don't match the Package Versioning Policy: http://www.haskell.org/haskellwiki/Package_versioning_policy Too bad that there's a policy. Although I personally don't like versioning using major and minor numbers, I understand that it's necessary for something like the haskell platform. Thanks for pointing me to it. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux)

iEYEARECAAYFAk65hzcACgkQtVwvsA+W4CClYwCfTw3+iq/I5MsZ1j+Zpa9d3icI 39YAnAsL2edv1TO4zE9qGkVy8dg/CD64 =L2jE -----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 08.11.2011 22:35, Rogan Creswick wrote:
> On Tue, Nov 8, 2011 at 11:47 AM, Jochen Keil
> 
wrote:
>>> Except we can't use these types of version "numbers" for
>>> Hackage, and they don't match the Package Versioning Policy: 
>>> http://www.haskell.org/haskellwiki/Package_versioning_policy
>> 
>> Too bad that there's a policy. Although I personally don't like 
>> versioning using major and minor numbers, I understand that it's 
>> necessary for something like the haskell platform. Thanks for
>> pointing me to it.
> 
> The PVP provides the framework that allows flexible version 
> specifications for dependencies.
> 
> Assuming the spirit of the PVP is followed, you can ignore the
> D-level version numbering, and depend on C-level numbering in a
> cabal file. That generally allows bugfix releases of your
> dependencies to be used without any changes to your code or cabal
> file, while telling cabal that API-breaking changes (marked by the
> higher version number slots) do not satisfy your application's
> requirements.
> 
> This is far from foolproof, but it is very convenient when it's
followed.
Sounds like libtool versioning to me, but I'm not the one to ask here.
- From a user (meaning non-programmer) perspective this is a bit odd. As
long as you don't break your API you are stuck with a possibly low major
version number which looks like alpha or beta software.
However, I understand the technical benefit of this and I'm fine with
it. Hope there are no marketing people around. :)
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)

iEYEARECAAYFAk65pcwACgkQtVwvsA+W4CC04ACfVfS9yLtRxTW/2PKrH7rHCzoi
E4wAn2yM97Oude2/WZE9e5l/Vgr1ywrL
=W/JC
-----END PGP SIGNATURE-----

On 11/07/2011 12:13 PM, Norbert Zeh wrote:

Let me address the two separate points here:

(1) No official release in 2 years: This makes me a bit uneasy about the future of xmonad, as it seems the original authors have lost interest in the project.

"Me, too". But it seems that the torch has been passed, not dropped. Truly, the state that no current coders have commit is a little skeery.

Bottom line: I highly recommend running the darcs version, particularly given the current state of affairs regarding point (1).

Understood, and I'm willing to go do that. But I think it's reasonable to point out that, in effect, xmonad positively repels new converts. As in, aggressively discourages them. Here's the path I've trod so far: - install vendor packages. Understand that I'll need to dog-paddle in haskell to function in xmonad. Tinker. Get frustrated - go learn about cabal, because bluetile suggests that it's a better way to get up-to-date code. On the way to bluetile, get enough xmonad functioning to not need it. Think "Ah, OK. vendor packages out of date, use the CPAN-equivalent for haskell, and all will be good". - Find that the haskell platform hasn't been updated in 7 months, despite the claim it'd get love in July. Get nervous about the entire haskell community. - See that cabal xmonad is -2 years!- out of date. Shudder. Understand that I need to learn darcs. ...... So, in order to use your WM, I have to completely bathe in haskell, down to learning the ideosyncratic, haskell-only VC you guys have cooked up. I _really like_ xmonad. I think it's a delightfully featureful and spare WM universe, and I'm feeling in better control of my desktop than I've been since I ditched CTWM. But insisting that new users be baptized into your church (or at least attend sunday school for a few months) is really bad for adoption. - Allen S. Rout

Excerpts from Allen S. Rout's message of Mon Nov 07 10:59:59 -0700 2011:

...

But I think it's reasonable to point out that, in effect, xmonad positively repels new converts. As in, aggressively discourages them.

Here's the path I've trod so far:

- install vendor packages. Understand that I'll need to dog-paddle in haskell to function in xmonad. Tinker. Get frustrated

- go learn about cabal, because bluetile suggests that it's a better way to get up-to-date code. On the way to bluetile, get enough xmonad functioning to not need it. Think "Ah, OK. vendor packages out of date, use the CPAN-equivalent for haskell, and all will be good".

- Find that the haskell platform hasn't been updated in 7 months, despite the claim it'd get love in July. Get nervous about the entire haskell community.

- See that cabal xmonad is -2 years!- out of date. Shudder. Understand that I need to learn darcs.

Point taken; your experience could have encouraged you to throw up your hands in disgust and give up at several points. Aside from our needing to fix the broken release schedules, I'm seeing that we also need to address documentation and communication problems: The installation instructions on xmonad.org should be much clearer for people who know nothing about and don't care about the haskell ecosystem, and just want to use xmonad. In reality you don't need to learn darcs. (See below re. installing darcs version.) You do need to install darcs, of course, which is problematic or frustrating for some. In my experience installing using darcs and cabal-install is much smoother and simpler than autotools based installs.

So, in order to use your WM, I have to completely bathe in haskell, down to learning the ideosyncratic, haskell-only VC you guys have cooked up.

It's really not very difficult to install from darcs, although it's not spelled out at all well on xmonad.org. The only darcs command you need is `darcs get': Installation Steps * Install ghc, darcs, and if available, cabal-install through your distro. Any combo of versions from the past 4 years or so that your distro has decided are compatible with one another are fine. * If cabal-īnstall isn't available in your repos, follow the simple instructions linked from the installation page: http://www.haskell.org/haskellwiki/Cabal-Install#Unix * In a convenienct directory like ~/repos or the like, run the following commands: $ darcs get http://code.haskell.org/xmonad $ darcs get http://code.haskell.org/XMonadContrib http://www.haskell.org/haskellwiki/Cabal-Install#Unix * Ensure that the C X11 libraries are available on your machine if you don't have them already. ( i.e. libXinerama libXext libX11 -- libx11-dev on debian based distros). * Then build and install xmonad and xmonad-contrib with the commands: $ cd $reponame $ cabal install

But insisting that new users be baptized into your church (or at least attend sunday school for a few months) is really bad for adoption.

What would you suggest to improve this situation? My experience with the xmonad community is that it's pretty friendly and responsive, so any suggestions you share should at least get a fair hearing. Best regards, -wmw

There seems to be a persistent notion that the current state of xmonad is something we actively encourage or have set as policy. It's not. We're as frustrated as you are, and we've been trying to get this sorted for some 2 years. What you're seeing is ugly workarounds, not something anyone involved with xmonad intends. (I should volunteer to do the release; it'd probably result in my immediately getting a job and being too busy to deal for the next 6 months....) [gallows humor?] -- brandon s allbery allbery.b@gmail.com wandering unix systems administrator (available) (412) 475-9364 vm/sms

Quoting "Allen S. Rout" :

darcs is a _D_vcs, right?

So if The Community has wanted a new release for two years, then what is preventing someone from saying "FINE! I call 0.99, _here_"?

Is there some official possession of credentials that is at issue?

Quoting "Jochen Keil" :

Sure, but why isn't it happening? :) What's keeping everyone from releasing the long debated (there were some mails a few weeks ago..) 1.0 version? I mean it can't be too hard to tag 1.0 in darcs and build a tarball.

The current de-facto maintainer seems to be Adam Vogt (aavogt in #xmonad). He told me yesterday that he's ready to make a release; the only holdup right now is that he does not have write access to xmonad.org nor control of the xmonad Twitter feed. The latter seems unimportant to me, but making a release and not publishing it on xmonad.org seems suboptimal. I've contacted the previous maintainer on Adam's behalf requesting control of the website, but haven't gotten a response yet. ~d

Excerpts from Anthony Campbell's message of Tue Nov 08 09:04:20 -0700 2011:

On 07 Nov 2011, Wirt Wolff wrote:

...

Excerpts from Allen S. Rout's message of Mon Nov 07 10:59:59 -0700 2011:

...

...

[snip]

...

Installation Steps [snip]

-wmw

I tried this, but got "There is no installed version of base".

I found some solutions offered for this on the web but none of them worked. I then reinstalled all the packages (Debian Sid) but still the same error.

Anthony

Hmm... base is part of ghc. Please pastebin or attach the output of `ghc-pkg list' and `ghc-pkg check' and we can help troubleshoot. Also, on #xmonad@freenode IRC people are often to help without requiring too long a lurking period after asking a question. So much for installation being smooth and easy for everyone. -wmw

On 9 November 2011 03:52, Anthony Campbell wrote:

On 08 Nov 2011, Wirt Wolff wrote:

...

...

I tried this, but got "There is no installed version of base".

I found some solutions offered for this on the web but none of them worked. I then reinstalled all the packages (Debian Sid) but still the same error.

Anthony

Hmm... base is part of ghc. Please pastebin or attach the output of `ghc-pkg list' and `ghc-pkg check' and we can help troubleshoot.

Also, on #xmonad@freenode IRC people are often to help without requiring too long a lurking period after asking a question.

So much for installation being smooth and easy for everyone.

-wmw

ghc-pkg gheck: no output

arcadia:~:$ ghc-pkg list /var/lib/ghc/package.conf.d                        Cabal-1.10.2.0   X11-1.5.0.0   X11-xft-0.3   array-0.3.0.2   base-4.3.1.0   bin-package-db-0.0.0.0   bytestring-0.9.1.10   containers-0.4.0.0   directory-1.1.0.0   extensible-exceptions-0.1.1.2   filepath-1.2.0.0   ghc-7.0.4   ghc-binary-0.5.0.2   ghc-prim-0.2.0.0   haskell2010-1.0.0.0   haskell98-1.1.0.1   hpc-0.5.0.6   integer-gmp-0.2.0.3   mtl-2.0.1.0   old-locale-1.0.0.2   old-time-1.0.0.6   pretty-1.0.1.2   process-1.0.1.5   random-1.0.0.3   rts-1.0   syb-0.3.3   template-haskell-2.5.0.0   time-1.2.0.3   transformers-0.2.2.0   unix-2.4.2.0   utf8-string-0.3.6   xmonad-0.9.2   xmonad-contrib-0.9.2

Has darcs xmonad somehow managed to be updated to _only_ work with 7.2.1? because it certainly appears that base-4 is installed there. -- Ivan Lazar Miljenovic Ivan.Miljenovic@gmail.com IvanMiljenovic.wordpress.com

Quoting "Allen S. Rout" :

...

$ darcs get http://code.haskell.org/XMonadContrib

Got the 'hash' message too, and also:

Thanks for the reminder. It looks like community.haskell.org has upgraded their darcs version, so we can finally fix this problem. I don't have commit bit on the xmonad repo, but I do have commit bit on the xmonad-contrib repo. Unless I hear complaints in the next, say, week, I'll upgrade the xmonad-contrib repo. I'm marking it on my calendar for November 16th now.

Warning: CRC errors found. These are probably harmless but should be repaired. See 'darcs gzcrcs --help' for more information.

Fixed. Again, thanks for the reminder. Everyone who cares about this warning should now run darcs gzcrcs --repair in their local XMonadContrib repository. You should not ever need to do this again, nor should new users who get a fresh copy of the repository (starting today) need to do this. ~d

c.h.o has upgraded darcs to 2.5 since that thread. ~d Quoting Gwern Branwen :

On Wed, Nov 9, 2011 at 3:48 PM, Allen S. Rout wrote:

...

Fetching a hashed repository would be faster.  Perhaps you could persuade the maintainer to run darcs optimize --upgrade with darcs 2.4.0 or higher?

As would one with checkpoints. I suggested the upgrade, along with releasing 1.0, long ago: http://www.reddit.com/r/xmonad/comments/casq8/proposal_upgrading_xmonad_darc...

-- gwern http://www.gwern.net

_______________________________________________ xmonad mailing list xmonad@haskell.org http://www.haskell.org/mailman/listinfo/xmonad

* On Wednesday, November 09 2011, wagnerdm@seas.upenn.edu wrote:

c.h.o has upgraded darcs to 2.5 since that thread. ~d

I've upgraded to the hashed format for both repos. In the unlikely case the old format repos are still useful archives of them can be found: http://code.haskell.org/~aavogt/xmonad_repo_old/ -- Adam

Hi, Am Dienstag, den 08.11.2011, 10:10 +1100 schrieb Ivan Lazar Miljenovic:

...

- Find that the haskell platform hasn't been updated in 7 months, despite the claim it'd get love in July. Get nervous about the entire haskell community.

I'm not sure why so many people make a fuss about the Haskell Platform not having a release... I can understand not _communicating_ this by updating the website, but is it really that big a deal (especially as there's no reasons for Linux distro packagers to stick solely to Haskell Platform releases)?

there is for those distros that aim to provide one version per package. Either we abandon the platform (but then why was it created in the first place) or we stick to the old versions, but then building some newer packages might not work. But this thread is getting O(ff)T here and more O(n)T on haskell-cafe... Greetings, Joachim -- Joachim "nomeata" Breitner mail@joachim-breitner.de | nomeata@debian.org | GPG: 0x4743206C xmpp: nomeata@joachim-breitner.de | http://www.joachim-breitner.de/

Last on this: On 11/07/2011 06:10 PM, Ivan Lazar Miljenovic wrote:

On 8 November 2011 04:59, Allen S. Rout wrote:

I'm not sure why so many people make a fuss about the Haskell Platform not having a release... I can understand not _communicating_ this by updating the website, but is it really that big a deal (especially as there's no reasons for Linux distro packagers to stick solely to Haskell Platform releases)?

Haskell Platform appears to be the face of haskell. This appearance could be wrong, but it's certainly the most well-designed and professional appearing haskell page I'd seen. So: claiming a release, then not hitting it, then not even getting around to retracting the claim, for months on end. If that's the peak performance of the haskell bunch putting on their best PR face, then that suggests the community is moribund. From that PR look, there's no way to distinguish between that impression and (say) a thriving collection of frequently maintained codebases. The latter seems closer to correct. That's why people care about it. - Allen S. Rout

On 11/08/2011 10:12 AM, Brandon Allbery wrote:

I'd kinda like this to be about getting xmonad releases back on track, and not a general b***hfest.

Right; topical, will do. :) - Allen S. Rout

On Mon, 07 Nov 2011 12:59:59 -0500,"Allen S. Rout" wrote:

On 11/07/2011 12:13 PM, Norbert Zeh wrote:

...

Let me address the two separate points here:

(1) No official release in 2 years: This makes me a bit uneasy about the future of xmonad, as it seems the original authors have lost interest in the project.

"Me, too". But it seems that the torch has been passed, not dropped. Truly, the state that no current coders have commit is a little skeery.

...

Bottom line: I highly recommend running the darcs version, particularly given the current state of affairs regarding point (1).

Understood, and I'm willing to go do that.

But I think it's reasonable to point out that, in effect, xmonad positively repels new converts. As in, aggressively discourages them.

Here's the path I've trod so far:

- install vendor packages. Understand that I'll need to dog-paddle in haskell to function in xmonad. Tinker. Get frustrated

- go learn about cabal, because bluetile suggests that it's a better way to get up-to-date code. On the way to bluetile, get enough xmonad functioning to not need it. Think "Ah, OK. vendor packages out of date, use the CPAN-equivalent for haskell, and all will be good".

- Find that the haskell platform hasn't been updated in 7 months, despite the claim it'd get love in July. Get nervous about the entire haskell community.

- See that cabal xmonad is -2 years!- out of date. Shudder. Understand that I need to learn darcs.

......

So, in order to use your WM, I have to completely bathe in haskell, down to learning the ideosyncratic, haskell-only VC you guys have cooked up.

I _really like_ xmonad. I think it's a delightfully featureful and spare WM universe, and I'm feeling in better control of my desktop than I've been since I ditched CTWM.

But insisting that new users be baptized into your church (or at least attend sunday school for a few months) is really bad for adoption.

I want to (partially) second the comments by Allen Rout. I started using Xmonad about 3 years ago, using the stock Debian package (by then, version 0.8.? ?). I do program, but not in Haskell, and I had tried other tiling wm before, including awesome, dwm, wmii, and ion3. Nevertheless, I was able to get xmonad to do (exactly?) what I wanted, given the amount of excellent documentation and tutorials. However, this took some time. Nevertheless, the end was really worth it. xmonad is the only WM I've used in the last 3 years, and is one of the first things I get going in any new machine. But now, when I read that 0.9 is obsolete (my current version, running Debian, is 0.9.2-2) and that I should grab the one in darcs, etc, etc, and possibly start fiddling around... well, I really don't feel tempted at all to try that right now (and I will, at least for now, feel solace in the "if it ain't broken, don't fix it"). Best, R.

_______________________________________________ xmonad mailing list xmonad@haskell.org http://www.haskell.org/mailman/listinfo/xmonad -- Ramon Diaz-Uriarte Department of Biochemistry, Lab B-25. Facultad de Medicina (UAM) Arzobispo Morcillo, 2 28029 Madrid Spain

Phone: +34-91-497-2412 Email: rdiaz02@gmail.com ramon.diaz@iib.uam.es http://ligarto.org/rdiaz

Jonas, thanks. You were right. It was straightforward (Debian includes cabal-install) and I did not notice any change (which is good!) after xmonad --recompile; xmonad --restart. Best, R. - On Tue, 8 Nov 2011 15:52:21 +0100,Jonas Bygdén wrote:

Ramon,

On Tue, Nov 8, 2011 at 15:27, Ramon Diaz-Uriarte wrote:

...

But now, when I read that 0.9 is obsolete (my current version, running Debian, is 0.9.2-2) and that I should grab the one in darcs, etc, etc, and possibly start fiddling around... well, I really don't feel tempted at all to try that right now (and I will, at least for now, feel solace in the "if it ain't broken, don't fix it").

I've followed the discussion for quite a while now, and realized that my "stock" xmonad (as delivered from ubuntu) has been outdated for years. I've felt the same way (i.e. "if it ain't broke, don't try to fix it") because it's just worked. I'm "just" a sysadm, and not a programmer, so I feel (as others have pointed out in this thread) more than a little intimidated by haskell.

Today however, with the fantastically simple explanation from Wirt Wolff earlier in this thread on how to build it with cabal from darcs, I decided to give it a go.

It really was as simple as Wirt describes. With a minimum of modification to my setup (mv /usr/bin/xmonad /usr/bin/.xmonad;ln -s /home/jonas/.cabal/bin/xmonad /usr/bin/xmonad), I'm now running the "latest and greatest" from darcs - integrated with gnome3http://haskell.org/haskellwiki/Xmonad/Using_xmonad_in_Gnome#Gnome3 and with xmonad-log-applet integrated in gnome3-panel.

It was a lot easier than I had expected.

Give it a go, it's easy to revert (just undo the rename and link) in case there's problems.

/Jonas -- Ramon Diaz-Uriarte Department of Biochemistry, Lab B-25. Facultad de Medicina (UAM) Arzobispo Morcillo, 2 28029 Madrid Spain

Phone: +34-91-497-2412 Email: rdiaz02@gmail.com ramon.diaz@iib.uam.es http://ligarto.org/rdiaz

On Wed, 9 Nov 2011 08:23:20 +0000,Anthony Campbell wrote:

On 08 Nov 2011, Ramon Diaz-Uriarte wrote:

...

Jonas, thanks. You were right. It was straightforward (Debian includes cabal-install) and I did not notice any change (which is good!) after xmonad --recompile; xmonad --restart.

Best,

R.

Doesn't work here (Debian Sid). It complains that there is no installed version of base. (See my earlier post about this on the list,)

Anthony, I am also using Debian (a mix of unstable and testing, with a few experiemental packages). I checked the output of your ghc-pkg list and it seems similar to mine. Both do include base-4.3.1.0. I really don't know how cabal works, but if you want, I can provide the output of any dpkg command. R.

-- Anthony Campbell - ac@acampbell.org.uk Microsoft-free zone - Using Debian GNU/Linux http://www.acampbell.org.uk - sample my ebooks at http://www.smashwords.com/profile/view/acampbell

_______________________________________________ xmonad mailing list xmonad@haskell.org http://www.haskell.org/mailman/listinfo/xmonad -- Ramon Diaz-Uriarte Department of Biochemistry, Lab B-25. Facultad de Medicina (UAM) Arzobispo Morcillo, 2 28029 Madrid Spain

Phone: +34-91-497-2412 Email: rdiaz02@gmail.com ramon.diaz@iib.uam.es http://ligarto.org/rdiaz

On 09 Nov 2011, Brandon Allbery wrote:

On Wed, Nov 9, 2011 at 08:21, Ramon Diaz-Uriarte <[1]rdiaz02@gmail.com> wrote:

On Wed, 9 Nov 2011 08:23:20 +0000,Anthony Campbell <[2]ac@acampbell.org.uk> wrote:

...

On 08 Nov 2011, Ramon Diaz-Uriarte wrote:

...

Jonas, thanks. You were right. It was straightforward (Debian includes cabal-install) and I did not notice any change (which is good!) after xmonad --recompile; xmonad --restart. Doesn't work here (Debian Sid). It complains that there is no installed version of base. (See my earlier post about this on the list,)

Anthony, I am also using Debian (a mix of unstable and testing, with a few experiemental packages). I checked the output of your ghc-pkg list and it seems similar to mine. Both do include base-4.3.1.0.

I'll have to dig back into this thread to find the mentioned post, but I will note that when cabal complains about not being able to satisfy "base -any" it *usually* means that there are conflicting boot library versions around that break the "base" package. � (That is, it really means that there is no combination of package versions that results in "base" being usable.) (ghc 7.2.x changed things in a way that can cause that error, by making the haskell98 package incompatible with modules compiled *without* haskell98; this breaks some of XMonadContrib's dependencies, in particular off the top of my head X11-xft uses haskell98 but XMC doesn't.) BTW, it would help if you posted the entire error you got.

Thanks to both for replies. There isn't much in the way of error messages: arcadia:~/xmonad-compile:$ cabal install Resolving dependencies... cabal: There is no installed version of base arcadia:~/xmonad-compile:$ That's it. Quite a few people on Google report this error but no solutions seem to exist. AC -- Anthony Campbell - ac@acampbell.org.uk Microsoft-free zone - Using Debian GNU/Linux http://www.acampbell.org.uk - sample my ebooks at http://www.smashwords.com/profile/view/acampbell

On 09 Nov 2011, Rogan Creswick wrote:

On Wed, Nov 9, 2011 at 9:27 AM, Anthony Campbell wrote:

...

Thanks to both for replies. There isn't much in the way of error messages:

arcadia:~/xmonad-compile:$ cabal install Resolving dependencies... cabal: There is no installed version of base arcadia:~/xmonad-compile:$

Try it with cabal-dev:

$ cabal install cabal-dev $ cd ~/xmonad-compile $ cabal-dev install

If it works, then the problem is related to your ghc user package db, if it fails, then look at the global package db and see if that looks wrong. If that looks right, then something weird is going on, but I suspect the previous steps will shed more light on the problem.

(It's possible you'll have a similar issue building cabal-dev, but let's hope not, at least for the moment.)

--Rogan

It seemed to work as far as the last step, with a lot of downloading and compiling. But cabal-dev didn't work until I located it in ~/.cabal/bin/. But again it complains about base: arcadia:~/xmonad-compile:$ ~/.cabal/bin/cabal-dev install Resolving dependencies... cabal: There is no installed version of base arcadia:~/xmonad-compile:$ Probably time to give up, I think. Anthony -- Anthony Campbell - ac@acampbell.org.uk Microsoft-free zone - Using Debian GNU/Linux http://www.acampbell.org.uk - sample my ebooks at http://www.smashwords.com/profile/view/acampbell