bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread pelzflorian (Florian Pelz)
On Sun, Apr 28, 2019 at 11:30:05PM -0500, Brett Gilio wrote:
> Hey all
> 
> I just reconfigured my system configuration after the big staging merge
> to master. I use StumpWM.
> 
> GDM seems to be starting just fine, but after this reconfiguration
> stumpwm or X or something is crashing for me and looping me back to GDM.
> 
> Where are the logs I can view for this? I see some message appear but
> quickly disappear before looping back into GDM.
> 
> If I can get the logs I can share them or try and figure it out myself.
> 
> Best,
> Brett Gilio
> 
> 
> 

I remember having a Gjs out of memory error in
/var/log/gdm/greeter.log before when logging in to GNOME.  Yours is
not an out of memory, is it?





bug#35475: [bug #35475] [PATCH] Add filesystem flag NOATIME

2019-04-29 Thread Ludovic Courtès
Hello,

rendaw <7e9wc56emja...@s.rendaw.me> skribis:

> From e7dcba4a76d624470d05172390dde8151fdb4750 Mon Sep 17 00:00:00 2001
> From: rendaw <>
> Date: Mon, 29 Apr 2019 01:00:17 +0900
> Subject: [PATCH] [bug #35475] [PATCH] Add filesystem flag NOATIME
>
> Add a no-atime flag to the file-system flags parameter and updates the 
> relevant documentation.
> ---
>  doc/guix.texi  | 11 +++
>  gnu/build/file-systems.scm |  2 ++
>  guix/build/syscalls.scm|  2 ++
>  3 files changed, 11 insertions(+), 4 deletions(-)

I tweaked the doc, added a commit log that follows our conventions¹, and
committed.

Thank you!

Ludo’.

¹ https://www.gnu.org/software/guix/manual/en/html_node/Submitting-Patches.html





bug#35475: [bug #35475] [PATCH] Add filesystem flag NOATIME

2019-04-29 Thread Danny Milosavljevic
Let's also add MS_RELATIME which is halfway between noatime and atime.

https://blog.confirm.ch/mount-options-atime-vs-relatime/


pgpuGHE4zyc7s.pgp
Description: OpenPGP digital signature


bug#35475: [bug #35475] [PATCH] Add filesystem flag NOATIME

2019-04-29 Thread Ludovic Courtès
Danny Milosavljevic  skribis:

> Let's also add MS_RELATIME which is halfway between noatime and atime.
>
> https://blog.confirm.ch/mount-options-atime-vs-relatime/

Good idea, could you send a patch?

Thanks,
Ludo’.





bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Timothy Sample
Hi Brett,

Brett Gilio  writes:

> Hey all
>
> I just reconfigured my system configuration after the big staging merge
> to master. I use StumpWM.
>
> GDM seems to be starting just fine, but after this reconfiguration
> stumpwm or X or something is crashing for me and looping me back to GDM.
>
> Where are the logs I can view for this? I see some message appear but
> quickly disappear before looping back into GDM.

GDM logs things to “/var/log/messages”.  However, it may not be useful
in this case (do take a look, though).

> If I can get the logs I can share them or try and figure it out myself.

You could also share some other details about your config.  For
instance, how is StumpWM started?  Do you have a “~/.xsession” script or
does StumpWM provide a “.desktop” file that GDM needs to find?


-- Tim





bug#35465: Ocaml-yojson build failing

2019-04-29 Thread Julien Lepiller
Le Sun, 28 Apr 2019 22:15:33 -0500,
Brett Gilio  a écrit :

> Julien Lepiller writes:
> >
> > Most likely this is my fault... I thought I rebuilt everything
> > locally though. Yojson can be upgraded to 1.7.0, but I think it
> > caused a build failure in merlin. Can you try?  
> 
> I actually did give this a try in my channel before sending the
> email. You are right, something with Merlin breaks because of Menhir.
> 
> I can get menhir to build just fine, but adding it as an input to
> Merlin doesnt change the behavior of the compile-time stuff.
> 
> So it is a rabbit hole.
> 
> If Ricardo is reading this, would you mind taking a look at this
> issue?
> 
> Brett

Fixed in 373a2a07975ed8e187f3baaa5c795a7ff0fd155b

Thanks for reporting this :)





bug#34203: scmutils fails to build

2019-04-29 Thread Ricardo Wurmus



> scmutils fails to build with the current MIT Scheme.

I have removed the scmutils package with commit 4267481a4e.

-- 
Ricardo






bug#35465: Ocaml-yojson build failing

2019-04-29 Thread Ricardo Wurmus


Brett Gilio  writes:

> If Ricardo is reading this, would you mind taking a look at this issue?

I’m reading this but I don’t know anything about Merlin or Menhir or
Ocaml-yojson.  I’m happy Julien fixed it already.

--
Ricardo






bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Ricardo Wurmus


Hey Brett,

> GDM seems to be starting just fine, but after this reconfiguration
> stumpwm or X or something is crashing for me and looping me back to GDM.
>
> Where are the logs I can view for this? I see some message appear but
> quickly disappear before looping back into GDM.

The gdm user’s home directory is /var/lib/gdm/; you may find some log
files there.

--
Ricardo






bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Timothy Sample
Hi again,

Timothy Sample  writes:

> Hi Brett,
>
> Brett Gilio  writes:
>
>> Hey all
>>
>> I just reconfigured my system configuration after the big staging merge
>> to master. I use StumpWM.
>>
>> GDM seems to be starting just fine, but after this reconfiguration
>> stumpwm or X or something is crashing for me and looping me back to GDM.
>>
>> Where are the logs I can view for this? I see some message appear but
>> quickly disappear before looping back into GDM.
>
> GDM logs things to “/var/log/messages”.  However, it may not be useful
> in this case (do take a look, though).
>
>> If I can get the logs I can share them or try and figure it out myself.
>
> You could also share some other details about your config.  For
> instance, how is StumpWM started?  Do you have a “~/.xsession” script or
> does StumpWM provide a “.desktop” file that GDM needs to find?

After doing some testing in a VM, it looks like this is an issue with my
recent commit: 8caa458953eeac783c73a0e5aaa72842fe3914c9.

I added a placeholder desktop entry file, and even though I did my best
to make it invisible, GDM is still selecting it.  (I tested GNOME and
XFCE, but I guess they were preferred by GDM over the placeholder,
whereas StumpWM is not.  Maybe the has to do with how the names are
sorted.)  I will try and fix this today, but in the mean time, you can
either:

• go back to using SLiM for now;
• launch StumpWM from “~/.xsession”; or
• install another WM, so that you can select StumpWM explicitly when
  logging in (using the little gear icon that shows up when you have
  multiple sessions available).

Sorry!


-- Tim





bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread brettg




On 29.04.2019 18:05, Timothy Sample wrote:

Hi again,

Timothy Sample  writes:


Hi Brett,

Brett Gilio  writes:


Hey all

I just reconfigured my system configuration after the big staging 
merge

to master. I use StumpWM.

GDM seems to be starting just fine, but after this reconfiguration
stumpwm or X or something is crashing for me and looping me back to 
GDM.


Where are the logs I can view for this? I see some message appear but
quickly disappear before looping back into GDM.


GDM logs things to “/var/log/messages”.  However, it may not be useful
in this case (do take a look, though).

If I can get the logs I can share them or try and figure it out 
myself.


You could also share some other details about your config.  For
instance, how is StumpWM started?  Do you have a “~/.xsession” script 
or

does StumpWM provide a “.desktop” file that GDM needs to find?


After doing some testing in a VM, it looks like this is an issue with 
my

recent commit: 8caa458953eeac783c73a0e5aaa72842fe3914c9.

I added a placeholder desktop entry file, and even though I did my best
to make it invisible, GDM is still selecting it.  (I tested GNOME and
XFCE, but I guess they were preferred by GDM over the placeholder,
whereas StumpWM is not.  Maybe the has to do with how the names are
sorted.)  I will try and fix this today, but in the mean time, you can
either:

• go back to using SLiM for now;
• launch StumpWM from “~/.xsession”; or
• install another WM, so that you can select StumpWM explicitly 
when
  logging in (using the little gear icon that shows up when you 
have

  multiple sessions available).

Sorry!


-- Tim


Thank you for looking into this Tim! I have gone back to SLiM for the 
time being until it is fixed :).


If anybody else is having this issue, going back to SLiM is really easy, 
check out my commit for reference.


https://github.com/brettgilio/guix-system/commit/64d389db13c2f78ee5c58af28c1639b098113c93





bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Timothy Sample
Hi Brett,

bre...@posteo.net writes:

> On 29.04.2019 18:05, Timothy Sample wrote:
>>
>> After doing some testing in a VM, it looks like this is an issue
>> with my
>> recent commit: 8caa458953eeac783c73a0e5aaa72842fe3914c9.
>>
>> I added a placeholder desktop entry file, and even though I did my best
>> to make it invisible, GDM is still selecting it.  (I tested GNOME and
>> XFCE, but I guess they were preferred by GDM over the placeholder,
>> whereas StumpWM is not.  Maybe the has to do with how the names are
>> sorted.)

This is exactly the problem.  To find a default session, it sorts the
names of all the “.desktop” files it can find (using “g_strcmp0”), and
picks the first.  Since we have “GNOME” < “XFCE” < “Fail” < “stumpwm”,
my tests did not catch this error.

I can think of two options for a fix before 1.0 (which is supposed to be
tomorrow!).  The cute one is to just rename “Fail” to “~Fail”, on the
expectation that this will come after most other names when sorted.  The
ugly one is to patch GDM to exclude the placeholder file when looking
for “.desktop” files, and then to select it instead of raising an error
when it can’t find anything.

My preference is for the ugly one, because the cute one feels like
putting a silly hack on top of silly hack – it’s just a bit too much.
I’ve attached a patch.  Thoughts?  (If I don’t hear anything, I will
push it – it’s important that this works for 1.0).

>From 6b78b1d9c9a4c5ecb55dcd973248310fbbea3a33 Mon Sep 17 00:00:00 2001
From: Timothy Sample 
Date: Mon, 29 Apr 2019 14:49:42 -0400
Subject: [PATCH] gnu: gdm: Avoid selecting the placeholder session.

This fixes .

* gnu/packages/gnome.scm (gdm)[arguments]: Modify the pre-configure phase to
patch GDM to only select the placeholder session if there are no others.
---
 gnu/packages/gnome.scm | 12 +++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/gnu/packages/gnome.scm b/gnu/packages/gnome.scm
index 17dc2fd102..f19b1836d3 100644
--- a/gnu/packages/gnome.scm
+++ b/gnu/packages/gnome.scm
@@ -5597,7 +5597,7 @@ libxml2.")
(modify-phases %standard-phases
  (add-before
   'configure 'pre-configure
-  (lambda* (#:key inputs #:allow-other-keys)
+  (lambda* (#:key inputs outputs #:allow-other-keys)
 ;; We don't have .
 (substitute* '("common/gdm-log.c"
"daemon/gdm-server.c"
@@ -5665,6 +5665,16 @@ libxml2.")
   (("\"gnome-session\"")
(string-append "\"" (assoc-ref inputs "gnome-session")
   "/bin/gnome-session\"")))
+;; Do not automatically select the placeholder session.
+(substitute* "daemon/gdm-session.c"
+  (("!g_str_has_suffix [(]base_name, \"\\.desktop\"[)]")
+   (string-append "!g_str_has_suffix (base_name, \".desktop\") || "
+  "(g_strcmp0(search_dirs[i], \""
+  (assoc-ref outputs "out") "/share/gdm/BuiltInSessions/"
+  "\") == 0 && "
+  "g_strcmp0(base_name, \"fail.desktop\") == 0)"))
+  (("g_error [(]\"GdmSession: no session desktop files installed, aborting\\.\\.\\.\"[)];")
+   "{ self->priv->fallback_session_name = g_strdup(\"fail\"); goto out; }"))
 #t))
  ;; GDM requires that there be at least one desktop entry
  ;; file.  This phase installs a hidden one that simply
-- 
2.21.0


In the future, we should find a way to make GDM errors less
catastrophic, but I doubt we could do that in a day (I certainly
couldn’t)!

> Thank you for looking into this Tim! I have gone back to SLiM for the
> time being until it is fixed :).
>
> If anybody else is having this issue, going back to SLiM is really
> easy, check out my commit for reference.
>
> https://github.com/brettgilio/guix-system/commit/64d389db13c2f78ee5c58af28c1639b098113c93

Thanks for providing this.  Hopefully it helps anybody else having
problems.


-- Tim


bug#34716: Guix test failure on i686: tests/processes

2019-04-29 Thread Ludovic Courtès
Hi Mark,

Mark H Weaver  skribis:

> test-name: not a client
> location: 
> /tmp/guix-build-guix-0.16.0-10.2637cfd.drv-0/source/tests/processes.scm:37
> source:
> + (test-assert
> +   "not a client"
> +   (not (find (lambda (session)
> +(= (getpid)
> +   (process-id (daemon-session-client session
> +  (daemon-sessions
> actual-value: #f
> actual-error:
> + (wrong-type-arg
> +   "struct_vtable"
> +   "Wrong type argument in position ~A (expecting ~A): ~S"
> +   (1 "struct" #f)
> +   (#f))
> result: FAIL

Fixed in commit c20d4cac1fb72afe26a3e534b92e9a9691936458.

Thanks,
Ludo’.





bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Ludovic Courtès
Hi Tim,

Timothy Sample  skribis:

> I can think of two options for a fix before 1.0 (which is supposed to be
> tomorrow!).  The cute one is to just rename “Fail” to “~Fail”, on the
> expectation that this will come after most other names when sorted.  The
> ugly one is to patch GDM to exclude the placeholder file when looking
> for “.desktop” files, and then to select it instead of raising an error
> when it can’t find anything.
>
> My preference is for the ugly one, because the cute one feels like
> putting a silly hack on top of silly hack – it’s just a bit too much.
> I’ve attached a patch.  Thoughts?  (If I don’t hear anything, I will
> push it – it’s important that this works for 1.0).

I don’t have an opinion as I didn’t follow this closely.

Could you make sure the fix works with the DEs and WMs that the
installer proposes (see (gnu installer services))?

If it does, could you push it also to the ‘version-1.0.0’ branch?

Thank you for looking into this!

Ludo’.





bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Brett Gilio


Timothy Sample writes:

> Hi Brett,
>
> bre...@posteo.net writes:
>
>> On 29.04.2019 18:05, Timothy Sample wrote:
>>>
>>> After doing some testing in a VM, it looks like this is an issue
>>> with my
>>> recent commit: 8caa458953eeac783c73a0e5aaa72842fe3914c9.
>>>
>>> I added a placeholder desktop entry file, and even though I did my best
>>> to make it invisible, GDM is still selecting it.  (I tested GNOME and
>>> XFCE, but I guess they were preferred by GDM over the placeholder,
>>> whereas StumpWM is not.  Maybe the has to do with how the names are
>>> sorted.)
>
> This is exactly the problem.  To find a default session, it sorts the
> names of all the “.desktop” files it can find (using “g_strcmp0”), and
> picks the first.  Since we have “GNOME” < “XFCE” < “Fail” < “stumpwm”,
> my tests did not catch this error.
>
> I can think of two options for a fix before 1.0 (which is supposed to be
> tomorrow!).  The cute one is to just rename “Fail” to “~Fail”, on the
> expectation that this will come after most other names when sorted.  The
> ugly one is to patch GDM to exclude the placeholder file when looking
> for “.desktop” files, and then to select it instead of raising an error
> when it can’t find anything.
>
> My preference is for the ugly one, because the cute one feels like
> putting a silly hack on top of silly hack – it’s just a bit too much.
> I’ve attached a patch.  Thoughts?  (If I don’t hear anything, I will
> push it – it’s important that this works for 1.0).
>
>
> In the future, we should find a way to make GDM errors less
> catastrophic, but I doubt we could do that in a day (I certainly
> couldn’t)!
>
>> Thank you for looking into this Tim! I have gone back to SLiM for the
>> time being until it is fixed :).
>>
>> If anybody else is having this issue, going back to SLiM is really
>> easy, check out my commit for reference.
>>
>> https://github.com/brettgilio/guix-system/commit/64d389db13c2f78ee5c58af28c1639b098113c93
>
> Thanks for providing this.  Hopefully it helps anybody else having
> problems.
>
>
> -- Tim

I think the uglier version is more generic and less likely to cause
future errors. But, it is a matter of time. The uglier one is likely
going to be more terse.

Do you need any help on my end?

Brett Gilio





bug#35500: ocaml-utop fatal error

2019-04-29 Thread Brett Gilio


I have ocaml-utop installed in an environment. When I run utop from the
shell it returns this rather unamusing error message.

brettg@guixsd ~ [env]$ utop
Fatal error: exception Not_found

There is a reference to this error on github, here
https://github.com/ocaml-community/utop/issues/224

I tried patching this myself but I did not get far.

Brett Gilio





bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Timothy Sample
Hi Ludo,

Ludovic Courtès  writes:

> Hi Tim,
>
> Timothy Sample  skribis:
>
>> I can think of two options for a fix before 1.0 (which is supposed to be
>> tomorrow!).  The cute one is to just rename “Fail” to “~Fail”, on the
>> expectation that this will come after most other names when sorted.  The
>> ugly one is to patch GDM to exclude the placeholder file when looking
>> for “.desktop” files, and then to select it instead of raising an error
>> when it can’t find anything.
>>
>> My preference is for the ugly one, because the cute one feels like
>> putting a silly hack on top of silly hack – it’s just a bit too much.
>> I’ve attached a patch.  Thoughts?  (If I don’t hear anything, I will
>> push it – it’s important that this works for 1.0).
>
> I don’t have an opinion as I didn’t follow this closely.

Okay.  I picked the more robust solution, but I still hope it doesn’t
stick around too long.  :)

> Could you make sure the fix works with the DEs and WMs that the
> installer proposes (see (gnu installer services))?

I tested all of the DEs listed there except for MATE, which I could not
compile.  I had to patch i3 and awesome to use absolute paths in their
“.desktop” files.  (The “xinitrc” script can find the binaries in $PATH,
but GDM cannot find the “TryExec” binaries unless they are absolute.
This is no doubt related to the other $PATH issues we saw with GDM.)

> If it does, could you push it also to the ‘version-1.0.0’ branch?

I pushed three commits to master:

1. e8c6e771c1 gnu: gdm: Avoid selecting the placeholder session.
2. 4ec42524dd gnu: awesome: Use absolute paths in session file.
3. aa7cdc57dc gnu: i3-wm: Use absolute paths in session files.

I then cherry-picked them onto the “version-1.0.0” branch.  Sorry if
that’s not what you meant.

> Thank you for looking into this!

No problem.  Thanks for the quick response!


-- Tim





bug#35484: GDM failing to start stumpwm after merge

2019-04-29 Thread Timothy Sample
Hi Brett,

Brett Gilio  writes:

> Timothy Sample writes:
>
>> Hi Brett,
>>
>> [...]
>>
>> I can think of two options for a fix before 1.0 (which is supposed to be
>> tomorrow!).  The cute one is to just rename “Fail” to “~Fail”, on the
>> expectation that this will come after most other names when sorted.  The
>> ugly one is to patch GDM to exclude the placeholder file when looking
>> for “.desktop” files, and then to select it instead of raising an error
>> when it can’t find anything.
>>
>> My preference is for the ugly one, because the cute one feels like
>> putting a silly hack on top of silly hack – it’s just a bit too much.
>> I’ve attached a patch.  Thoughts?  (If I don’t hear anything, I will
>> push it – it’s important that this works for 1.0).
>>
>>
>> In the future, we should find a way to make GDM errors less
>> catastrophic, but I doubt we could do that in a day (I certainly
>> couldn’t)!
>>
>>> Thank you for looking into this Tim! I have gone back to SLiM for the
>>> time being until it is fixed :).
>>>
>>> If anybody else is having this issue, going back to SLiM is really
>>> easy, check out my commit for reference.
>>>
>>> https://github.com/brettgilio/guix-system/commit/64d389db13c2f78ee5c58af28c1639b098113c93
>>
>> Thanks for providing this.  Hopefully it helps anybody else having
>> problems.
>>
>>
>> -- Tim
>
> I think the uglier version is more generic and less likely to cause
> future errors. But, it is a matter of time. The uglier one is likely
> going to be more terse.
>
> Do you need any help on my end?

No, but thank you for offering and thanks for catching my mistake before
we released 1.0.  :)

I pushed e8c6e771c199c02b748eb66d3a03c46d2a32ca29 to master, which
should fix this.  Hence, I’m closing the bug.


-- Tim





bug#35501: libplist 2.0.0 build failure: File does not exist.

2019-04-29 Thread Brendan Tildesley
=
   libplist 2.0.0: test/test-suite.log
=

# TOTAL: 29
# PASS:  26
# SKIP:  0
# XFAIL: 0
# FAIL:  3
# XPASS: 0
# ERROR: 0

.. contents:: :depth: 2

FAIL: largecmp
==

File does not exists
FAIL largecmp.test (exit status: 2)

FAIL: hugecmp
=

File does not exists
FAIL hugecmp.test (exit status: 2)

FAIL: bigarraycmp
=

File does not exists
FAIL bigarraycmp.test (exit status: 2)








bug#35475: [bug #35475] [PATCH] Add filesystem flag NOATIME

2019-04-29 Thread rendaw
On 4/29/19 7:14 PM, Ludovic Courtès wrote:
> Hello,
>
> rendaw <7e9wc56emja...@s.rendaw.me> skribis:
>
>> From e7dcba4a76d624470d05172390dde8151fdb4750 Mon Sep 17 00:00:00 2001
>> From: rendaw <>
>> Date: Mon, 29 Apr 2019 01:00:17 +0900
>> Subject: [PATCH] [bug #35475] [PATCH] Add filesystem flag NOATIME
>>
>> Add a no-atime flag to the file-system flags parameter and updates the 
>> relevant documentation.
>> ---
>>  doc/guix.texi  | 11 +++
>>  gnu/build/file-systems.scm |  2 ++
>>  guix/build/syscalls.scm|  2 ++
>>  3 files changed, 11 insertions(+), 4 deletions(-)
> I tweaked the doc, added a commit log that follows our conventions¹, and
> committed.
>
> Thank you!
>
> Ludo’.
>
> ¹ 
> https://www.gnu.org/software/guix/manual/en/html_node/Submitting-Patches.html

Thank you!  And sorry, I'll read more carefully next time.