[bug#72538] [PATCH] gnu: xmonad: Re-add xmonad.desktop file.

[bug#72538] [PATCH] gnu: xmonad: Re-add xmonad.desktop file.

[EuAndreh via Guix-patches via]

* gnu/packages/wm.scm (xmonad)[arguments]: Add 'install-xsession phase
  that was removed in previous automated update.

Change-Id: Iac02b4c6e7a73c9b354c52b46eb39dca8156b1c3
---

Notes:
    The said automated update is 0bce74d458a343e61d054c4b25d6f67bd1086f3c,
    an upgrade to Stackage 20.26.
    
    After rebooting my computer and realizing what happened, I searched for
    discussions around this on the mailing lists, only to find a message
    from, wait a minute, me! haha
    
    This has happened before, and I was the one to send an equivalent patch
    to address the issue, 307d1b626be86ed21d48d44a131ce8490f370a17.
    
    I did take a quick look at this latest automated update, and all I could
    find were docstring typos spelling "httsp", and not much else.  Other
    packages with custom phases were either adjusted or kept as they were,
    so it does seem to have been a one-off in this update.
    
    Cc Lars-Dominik Braun:
    If you're so inclined, I may help to improve the upgrade script to avoid
    this in the future, but also feel free to decline or ignore.  And thanks
    for the maintenance work on the Haskell packages :)

 gnu/packages/wm.scm | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/gnu/packages/wm.scm b/gnu/packages/wm.scm
index ad74fb5cf7..da1695ff62 100644
--- a/gnu/packages/wm.scm
+++ b/gnu/packages/wm.scm
@@ -872,6 +872,24 @@ (define-public xmonad
         (base32 "19qz9a23377nzc0qq8nca45s745mfncd4i2vwba14gi7ipipfcil"))))
     (build-system haskell-build-system)
     (properties '((upstream-name . "xmonad")))
+    (arguments
+      (list
+       #:phases
+       #~(modify-phases %standard-phases
+           (add-after 'install 'install-xsession
+             (lambda _
+               (let* ((xsessions (string-append #$output "/share/xsessions"))
+                      (entry     (string-append xsessions "/xmonad.desktop")))
+                 (mkdir-p xsessions)
+                 (call-with-output-file
+                  entry
+                  (lambda (port)
+                    (format port "~
+                      [Desktop Entry]~@
+                      Name=xmonad~@
+                      Comment=xmonad window manager~@
+                      Exec=~a/bin/xmonad~@
+                      Type=Application~%" #$output)))))))))
     (inputs (list ghc-x11 ghc-data-default-class ghc-setlocale))
     (native-inputs (list ghc-quickcheck ghc-quickcheck-classes))
     (home-page "http://xmonad.org")

base-commit: 9c6353c5568a56620af43c2213a9f1cbdc2b4adf
-- 
2.45.2

bug#72538: [PATCH] gnu: xmonad: Re-add xmonad.desktop file.

[Lars-Dominik Braun]

[-- Attachment #1: Type: text/plain, Size: 917 bytes --]

Hi,

merged as commit bc71034f789603349080c2f35570910c07d526ec. I also fixed
the typos I introduced.

>     This has happened before, and I was the one to send an equivalent patch
>     to address the issue, 307d1b626be86ed21d48d44a131ce8490f370a17.

I’m sorry :( My usual flow is to run my update script (which only
touches updated packages) and then manually revert unwanted changes,
but I did not catch the xmonad phase. Is there any way we can upstream
the .desktop file?

>     If you're so inclined, I may help to improve the upgrade script to avoid
>     this in the future, but also feel free to decline or ignore.  And thanks
>     for the maintenance work on the Haskell packages :)

See attached script. The thing is that we *want* to remove unnecessary
phases during package upgrades. So I don’t know how to differentiate
between wanted and unwanted changes, except for manually checking them.

Lars


[-- Attachment #2: haskell-update.scm --]
[-- Type: application/vnd.lotus-screencam, Size: 8819 bytes --]

[bug#72538] [PATCH] gnu: xmonad: Re-add xmonad.desktop file.

[EuAndreh via Guix-patches via]

Hi, thanks for the very quick response!

Upstreaming the change is indeed than either improving the upgrade 
script or keeping the ad-hoc phase around.  I hadn't thought of that. 
I'll do that eventually when I get it :)

Thanks for the maintenance work.