Re: Circular dependencies in python modules

[Federico Beffa]

Andreas Enge  writes:

> Looking at the code of both modules on github, there is:
> https://github.com/RDFLib/sparqlwrapper/blob/master/requirements.txt
>rdflib
>rdflib_jsonld
> and
> https://github.com/RDFLib/rdflib/blob/master/requirements.py2.txt
>flake8
>isodate
>pyparsing<=1.5.7
>SPARQLWrapper
> (and similarly for python3; no idea why there the installation worked
> without sparqlwrapper).
>
> So what can we do?

It appears that SPARQLWrapper is not really needed and you can disable
dependency guessing with '--no-guessing-deps'. See

https://sources.debian.net/src/rdflib/4.1.2-3/debian/rules/

Regards,
Fede

Re: A couple roadblocks using guix

[Adam Pribyl]

On Wed, 21 Jan 2015, Omar Radwan wrote:


I'm trying to install guix on my laptop(finally got the thinkpenguin wifi
card), after reading through the manual, I have everything setup, except
there are a couple bumps on the road.

One of the things is that during system initialization, the (gnu system
networking) module returns an error which reads "guix system: error: failed
to load operating system file '/mnt/etc/config.scm': (misc-error #f "~A ~S"
("no code for module" (gnu system networking)) #f)
and that kind of makes things really annoying to manually restart
networking every reboot.


Are you sure you included a proper module(s)?
(use-modules (gnu)
 (gnu services ssh)
 (gnu services networking)
 (gnu services xorg)
 )

(btw use guix system reconfigure /config.scm if you have already installed 
system)



Another thing is that is sort of a problem is during a windowmaker X
session, all my graphical applications(xterm, icecat, etc), are not listed
in the application menu and are not runnable, and I would think it has
something to do with user permissions, but an X session in root also
doesn't let me run them.


This I experince too. In my opinion the reason is, that applications are 
installed per user and the path to them is just in your ~/.guix-profile/. 
I am not sure how WM searches for them, but definitely not in 
~/.guix-profile, but I am not sure about this.



The last thing is that I have not found any meta-packages, which allow me
to install multiple packages that are needed for the same program, like for
instance, there is gcc, make, cmake, and all the bread-and-butter build
programs, but you have to name each in the "guix package -i *" declaration
to install, but if there was some metapackage, like buildessential (I come
from Debian).


That last peeve is really just a missing feature, I do understand that it
will take some time to implement and I'm not really demanding it, just
extremely recommending it. But the first 2 are definitely problems that I
really hope someone could help me with.



Adam Pribyl

[PATCH] gnu: sudo: Set rundir, vardir and iologdir.

[宋文武]

* gnu/packages/admin.scm (sudo)[arguments]<#:configure-flags>: Add
  --with-rundir=/run/sudo --with-vardir=/var/db/sudo
  --with-iologdir=/var/log/sudo-io.
---
 gnu/packages/admin.scm | 13 +++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/gnu/packages/admin.scm b/gnu/packages/admin.scm
index d7cd0b8..534fd07 100644
--- a/gnu/packages/admin.scm
+++ b/gnu/packages/admin.scm
@@ -631,7 +631,10 @@ system administrator.")
 "002l6h27pnhb77b65frhazbhknsxvrsnkpi43j7i0qw1lrgi7nkf"
 (build-system gnu-build-system)
 (arguments
- `(#:configure-flags '("--with-logpath=/var/log/sudo.log")
+ `(#:configure-flags '("--with-logpath=/var/log/sudo.log"
+   "--with-rundir=/run/sudo"
+   "--with-vardir=/var/db/sudo"
+   "--with-iologdir=/var/log/sudo-io")
#:phases (alist-cons-before
  'configure 'pre-configure
  (lambda _
@@ -645,7 +648,13 @@ system administrator.")
   "")
  (("^install: (.*)install-sudoers(.*)" _ before after)
   ;; Don't try to create /etc/sudoers.
-  (string-append "install: " before after "\n"
+  (string-append "install: " before after "\n"))
+ (("\\$\\(DESTDIR\\)\\$\\(rundir\\)")
+  ;; Don't try to create /run/sudo.
+  "$(TMPDIR)/dummy")
+ (("\\$\\(DESTDIR\\)\\$\\(vardir\\)")
+  ;; Don't try to create /var/db/sudo.
+  "$(TMPDIR)/dummy")))
  %standard-phases)
 
;; XXX: The 'testsudoers' test series expects user 'root' to exist, but
-- 
2.1.2

Re: [PATCH] gnu: sudo: Set rundir, vardir and iologdir.

[宋文武]

This make /ts -> /run/sudo/ts, /lectured -> /var/db/lectured.

Re: A couple roadblocks using guix

[Omar Radwan]

>Are you sure you included a proper module(s)?
>(use-modules (gnu)
>(gnu services ssh)
> (gnu services networking)
>(gnu services xorg)
 >   )

In the manual
https://www.gnu.org/software/guix/manual/guix.html#Networking-Services it
says The (gnu system networking) module provides services to configure the
network interface. Not (gnu services networking), I'll try that one though.


>This I experince too. In my opinion the reason is, that applications are
installed per user and the path to them is just in your ~/.guix->profile/.
I am not sure how WM searches for them, but definitely not in
~/.guix-profile, but I am not sure about this.

There must be a way around this since it probably searches in /bin and
/usr/bin, but their is a way to go around it if you could add the path
/gnu/store/*xterm(or whatever*/bin/xterm(or whatever

On Thu, Jan 22, 2015 at 5:02 AM, Adam Pribyl  wrote:

> On Wed, 21 Jan 2015, Omar Radwan wrote:
>
>  I'm trying to install guix on my laptop(finally got the thinkpenguin wifi
>> card), after reading through the manual, I have everything setup, except
>> there are a couple bumps on the road.
>>
>> One of the things is that during system initialization, the (gnu system
>> networking) module returns an error which reads "guix system: error:
>> failed
>> to load operating system file '/mnt/etc/config.scm': (misc-error #f "~A
>> ~S"
>> ("no code for module" (gnu system networking)) #f)
>> and that kind of makes things really annoying to manually restart
>> networking every reboot.
>>
>
> Are you sure you included a proper module(s)?
> (use-modules (gnu)
>  (gnu services ssh)
>  (gnu services networking)
>  (gnu services xorg)
>  )
>
> (btw use guix system reconfigure /config.scm if you have already installed
> system)
>
>  Another thing is that is sort of a problem is during a windowmaker X
>> session, all my graphical applications(xterm, icecat, etc), are not listed
>> in the application menu and are not runnable, and I would think it has
>> something to do with user permissions, but an X session in root also
>> doesn't let me run them.
>>
>
> This I experince too. In my opinion the reason is, that applications are
> installed per user and the path to them is just in your ~/.guix-profile/. I
> am not sure how WM searches for them, but definitely not in
> ~/.guix-profile, but I am not sure about this.
>
>
>  The last thing is that I have not found any meta-packages, which allow me
>> to install multiple packages that are needed for the same program, like
>> for
>> instance, there is gcc, make, cmake, and all the bread-and-butter build
>> programs, but you have to name each in the "guix package -i *" declaration
>> to install, but if there was some metapackage, like buildessential (I come
>> from Debian).
>>
>>
>> That last peeve is really just a missing feature, I do understand that it
>> will take some time to implement and I'm not really demanding it, just
>> extremely recommending it. But the first 2 are definitely problems that I
>> really hope someone could help me with.
>>
>>
> Adam Pribyl
>
>

Re: [PATCH] gnu: sudo: Set rundir, vardir and iologdir.

[Ludovic Courtès]

宋文武  skribis:

> * gnu/packages/admin.scm (sudo)[arguments]<#:configure-flags>: Add
>   --with-rundir=/run/sudo --with-vardir=/var/db/sudo
>   --with-iologdir=/var/log/sudo-io.

Definitely an improvement, please push!

Thanks,
Ludo’.

Re: [PATCH v2] gnu.scm: add syntatic sugar macros for use-modules of packages, services, and system in the OS EDSL.

[Ludovic Courtès]

jgr...@parenthetical.io skribis:

> From ee940d31ec9b3847c1bd9dfe817ba8f6550361bc Mon Sep 17 00:00:00 2001
> From: "Joshua S. Grant" 
> Date: Sun, 18 Jan 2015 16:23:21 -0600
> Subject: [PATCH] [PATCH] gnu.scm: add syntactic sugar macros for use-modules
>  of packages, services, and system in the OS EDSL.

Applied with a proper ChangeLog-style commit.   Please check ‘HACKING’
and previous commit for examples on how to do that.

Thanks!

Ludo’.

Re: [PATCH] gnu: Add most debian patches to nvi.

[Ludovic Courtès]

Marek Benc  skribis:

> On 01/21/2015 04:03 PM, Ludovic Courtès wrote:
>> I agree with Andreas: there are 20 patches, and that level of patching
>> goes beyond our mission, IMO.
>>
>> I would rather leave it as is if it’s usable, or remove it, or point to
>> a new upstream if there is one.
>
> In that case, I think at least two of the patches should be allowed to
> get in, the ones about the Berkely db library, since it's spewing out
> errors without them. (but the editor itself works)

You mean there are run-time errors without nvi-db4.patch?  What kind?

The second patch is about page sizes apparently.  Do you know what the
deal is?

Two/three patches would be OK, but we need to have an understanding of
why we’re carrying these particular patches.

Thank you!

Ludo’.

Re: [PATCH] gnu: Add most debian patches to nvi.

[Marek Benc]

On 01/22/2015 05:05 PM, Ludovic Courtès wrote:

Marek Benc  skribis:


On 01/21/2015 04:03 PM, Ludovic Courtès wrote:

I agree with Andreas: there are 20 patches, and that level of patching
goes beyond our mission, IMO.

I would rather leave it as is if it’s usable, or remove it, or point to
a new upstream if there is one.

In that case, I think at least two of the patches should be allowed to
get in, the ones about the Berkely db library, since it's spewing out
errors without them. (but the editor itself works)

You mean there are run-time errors without nvi-db4.patch?  What kind?


I mistyped, it's only warnings, the editor seems to work okay. In 
particular,

the warnings look like this:

BDB0635 DB_CREATE must be specified to create databases.
BDB0511 page sizes must be a power-of-2


The second patch is about page sizes apparently.  Do you know what the
deal is?


nvi seems to store the data of files in Berkely DB database objects, and 
when

setting up a database for a file, it determines a page size to fit the file
into 15 pages or less, using a page size of at least 1K and at most 10K.

The way it does the calculation, I quote:
  psize = ((sb.st_size / 15) + 1023) / 1024;
  if (psize > 10)
psize = 10;
  if (psize == 0)
psize = 1;
  psize *= 1024;

makes psize hold a variable that's a multiple of 1024, but not necessarilya
power of two (for example, 3072). This didn't use to be a problem 
before, but
nowadays, BDB requires pages to be powers of two, and so the modified 
code, I

quote:

  psize = ((sb.st_size / 15) + 1023) / 1024;
  if (psize >= 8) psize=8<<10;
  else if (psize >= 4) psize=4<<10;
  else if (psize >= 2) psize=2<<10;
  else psize=1<<10;
  psize *= 1024;

fixes that problem.


Two/three patches would be OK, but we need to have an understanding of
why we’re carrying these particular patches.


Sure, I understand.



Thank you!

Ludo’.


Marek

Re: [PATCH] gnu: Add most debian patches to nvi.

[Marek Benc]

On 01/22/2015 05:39 PM, Marek Benc wrote:

psize = ((sb.st_size / 15) + 1023) / 1024;
  if (psize >= 8) psize=8<<10;
  else if (psize >= 4) psize=4<<10;
  else if (psize >= 2) psize=2<<10;
  else psize=1<<10;
  psize *= 1024; 
Oops, ignore the last line, it's a typo on my end and not a problem of 
the patch (the patch removes it).

New year, new name²

[Ludovic Courtès]

It is my pleasure to announce the new new name to designate the
standalone distro: the Guix System Distribution, or GNU GSD.

I will update the web page and manual accordingly.

For those who haven’t followed, the discussion is at
.  
And yes, this was the last discussion on that topic.

Thanks to everyone who contributed!

Ludo’.


signature.asc
Description: PGP signature

Re: Large git repositories

[Andreas Enge]

Hello! Sorry for my late reply, I have been working on something else in the
meantime.

On Tue, Jan 13, 2015 at 01:31:51PM +, Amirouche Boubekki wrote:
> Google web fonts has droid font available in zip format via http [1],
> not sure if the link is works across updates.

Thanks, that could be interesting (assuming that zips do not contain time 
stamps).

On Thu, Jan 22, 2015 at 01:33:57PM +, Amirouche Boubekki wrote:
> This is again regarding the issue of downloading droid font in particular.
> This can be done via github with links like the following:
> https://github.com/android/platform_frameworks_base/raw/
> feef9887e8f8eb6f64fc1b4552c02efb5755cdc1/data/fonts/DroidSans.ttf

Nice, thanks! What does the "feef..." stand for? I assumed it would be the
git commit hash, but there is none of this form. The problem here is that
we need about 20 different files for the full family including different
scripts.

On Tue, Jan 13, 2015 at 11:32:25AM +0100, Ludovic Courtès wrote:
> Andreas Enge  skribis:
> > It looks like:
> >--depth=1 --single-branch
> > could be useful parameters for "git clone".
> Yes, that’s what came to mind.
> If it happens to be useful, we could extend ‘git-reference’ with a list
> of extra parameters to pass to ‘git’.

Actually, I have something else in mind. Currently, we accept commit hashes
or branch/tag names. How about adding an additional parameter "branch?"
(in various places, when I looked at the code I think I understood where
and would be willing to propose a patch)? If it is set, we can directly
specify the branch/tag at clone, without the need for an additional checkout,
and could also pass the parameters "--depth=1 --single-branch".

Alternatively, we could also automatically distinguish commit hashes from
branch/tag names, for instance, by imposing that always the long commit hash
is used and assuming that a branch/tag name never equals a hexidecimal string
of the corresponding length. But I prefer the additional parameter.

In our case, we could then git clone the 1.5 GB once and for all and delete
everything outside data/fonts (and in particular the huge .git). The result
would be the source for a variety of android fonts (maybe to be placed into
font-android.scm or the like).

What do you think?

Andreas

Re: Circular dependencies in python modules

[Andreas Enge]

On Thu, Jan 22, 2015 at 09:50:35AM +0100, Federico Beffa wrote:
> It appears that SPARQLWrapper is not really needed and you can disable
> dependency guessing with '--no-guessing-deps'. See
> https://sources.debian.net/src/rdflib/4.1.2-3/debian/rules/

Thanks for the suggestion! This seems to be some debian specific macro in
their package recipes; I tried it as a configure flag, and setup.py chokes
on it.

However, there are these lines in setup.py:
kwargs['install_requires'] = [
'isodate',
'pyparsing', 'SPARQLWrapper']
explicitly in the "else" branch of
  if sys.version_info[0] >= 3:
So this would explain why there was no problem with python-3.

I would suggest to patch this 'SPARQLWrapper' away. What is the recommended
way nowadays? A patch, I suppose? Or a snippet?

There is another occurrence of SPARQLWrapper in rdflib.egg-info/requires.txt.
But I suppose this is merely documentation?

Andreas

sockets availability during build

[Federico Beffa]

Hi,

I'm working on libupnp. The 'check phase gives the following error:

Initializing UPnP ...
** ERROR UpnpInit(): -208 UPNP_E_SOCKET_ERROR
FAIL: test_init

In spite of this, if I "guix environment" and run the test binary it succeeds:

Initializing UPnP ...
UPnP Initialized OK ip=192.168.0.5, port=49152

I suppose that sockets can't be accessed during the build phase. Is
that correct?

Thanks,
Fede

Re: sockets availability during build

[John Darrington]

On Thu, Jan 22, 2015 at 08:54:11PM +0100, Federico Beffa wrote:
 Hi,
 
 I'm working on libupnp. The 'check phase gives the following error:
 
 Initializing UPnP ...
 ** ERROR UpnpInit(): -208 UPNP_E_SOCKET_ERROR
 FAIL: test_init
 
 In spite of this, if I "guix environment" and run the test binary it 
succeeds:
 
 Initializing UPnP ...
 UPnP Initialized OK ip=192.168.0.5, port=49152
 
 I suppose that sockets can't be accessed during the build phase. Is
 that correct?

local sockets (AF_UNIX) can be.  But the only Internet sockets accessible are 
those on the local (lo) interface.
All other interfaces are disabled.

J'

-- 
PGP Public key ID: 1024D/2DE827B3 
fingerprint = 8797 A26D 0854 2EAB 0285  A290 8A67 719C 2DE8 27B3
See http://sks-keyservers.net or any PGP keyserver for public key.



signature.asc
Description: Digital signature

Re: [PATCH] gnu: libpeas: Update to 1.12.1, add arguments and disable tests.

[Ludovic Courtès]

Federico Beffa  skribis:

> This package, up to now, is the only one experiencing some
> side-effects from the gobject-introspection patch that we adopted from
> nix and which installs the full path name of shared libraries in
> .typelib and .gir files.
>
> The side-effect here is that some tests fail because:
>
> * they are executed before installation (solvable by swapping 'check
> and 'install phases) and

That’s not a very good sign though.  In general, it should be possible
to run tests before installation.

> * because it makes use of libtool convenience libraries (libraries
> which are only intermediate steps and are not installed). This is
> implemented by specifying a fake installation directory called
> "/nowhere" which, of course, does not exists. This is the error
> message:
>
> /tmp/nix-build-libpeas-1.12.1.drv-0/libpeas-1.12.1/tests/libpeas/.libs/extension-c:11955):
> libpeas-ERROR **: Unhandled warning: (null): Failed to load shared
> library '/nowhere/libintrospection-1.0.so.0' referenced by the
> typelib: /nowhere/libintrospection-1.0.so.0: cannot open shared object
> file: No such file or directory

Does it mean that it’s trying to dlopen a convenience library?  This
shouldn’t work anyway, right?

Thanks for investigating!

Ludo’.

Re: [PATCH] gnu: libpeas: Update to 1.12.1, add arguments and disable tests.

[Ludovic Courtès]

Federico Beffa  skribis:

> On Tue, Jan 20, 2015 at 6:52 PM, Federico Beffa  wrote:
>> Please find attached a patch for libpeas.
>>
>> This package, up to now, is the only one experiencing some
>> side-effects from the gobject-introspection patch that we adopted from
>> nix and which installs the full path name of shared libraries in
>> .typelib and .gir files.
>
> I find the above side-effect disturbing. For this reason I would like
> to propose an additional patch to 'gobject-introspection'. When
> 'gobject-introspection' consults a .typelib file (due to the patch we
> introduced earlier) it will find a reference to a shared library which
> includes the absolute path. With the new patch, if the library is not
> found at the absolute path, then it discards the path and tries to
> look for the library using the OS dynamic library loading
> infrastructure (which is the normal behavior of
> 'gobject-introspection').

I see, that makes sense indeed.

> With this change 'libpeas' (and all its dependencies) passes all tests.

Perfect!

> From 58d32305b609eccf54de620398a54457af0af0c5 Mon Sep 17 00:00:00 2001
> From: Federico Beffa 
> Date: Tue, 20 Jan 2015 18:43:55 +0100
> Subject: [PATCH 1/2] gnu: libpeas: Update to 1.12.1, add 'pre-build phase.
>
> * gnu/packages/gnome.scm (libpeas): Update to version 1.12.1 and add
>   'pre-build phase.

OK.

> From b90d70f312e9c298232d76f29099002f359f4468 Mon Sep 17 00:00:00 2001
> From: Federico Beffa 
> Date: Wed, 21 Jan 2015 13:46:52 +0100
> Subject: [PATCH 2/2] gnu: gobject-introspection: Add patch.
>
> * gnu/packages/glib.scm (gobject-introspection): Add patch
>   gobject-introspection-girepository.patch.

[...]

> --- /dev/null
> +++ b/gnu/packages/patches/gobject-introspection-girepository.patch
> @@ -0,0 +1,17 @@
> +--- gobject-introspection-1.42.0/girepository/gitypelib.c.orig   
> 2015-01-20 20:37:43.706684642 +0100
>  gobject-introspection-1.42.0/girepository/gitypelib.c2015-01-21 
> 09:19:40.927751124 +0100
> +@@ -2296,7 +2296,14 @@
> + {
> +   GModule *module;
> + 
> ++  /* We expect an absolute directory name.  If the library is
> ++ not found, try with just the basename and the system
> ++ dynamic library infrastructure. */
> +   module = load_one_shared_library (shlibs[i]);
> ++  if (module == NULL && g_path_is_absolute (shlibs[i]))
> ++{
> ++  module = load_one_shared_library (g_basename(shlibs[i]));

Please explain at the top of the paragraph that libgobject-introspection
normally looks for library in the dynamic linker’s search path, and that
because we use absolute file names in gir files, we need this patch to
get closer to the initial behavior.

OK to push with this change.

Thank you!

Ludo’.

Re: A couple roadblocks using guix

[Ludovic Courtès]

Omar Radwan  skribis:

> One of the things is that during system initialization, the (gnu system
> networking) module returns an error which reads "guix system: error: failed
> to load operating system file '/mnt/etc/config.scm': (misc-error #f "~A ~S"
> ("no code for module" (gnu system networking)) #f)
> and that kind of makes things really annoying to manually restart
> networking every reboot.

There’s a typo in the 0.8 manual (which was fixed in fa1e31b8): it
should read (gnu services networking).

> Another thing is that is sort of a problem is during a windowmaker X
> session, all my graphical applications(xterm, icecat, etc), are not listed
> in the application menu and are not runnable

I think this is the same problem as , which
was fixed a while back.

This suggests that it’s high time for a release.  ;-)

> The last thing is that I have not found any meta-packages, which allow me
> to install multiple packages that are needed for the same program, like for
> instance, there is gcc, make, cmake, and all the bread-and-butter build
> programs, but you have to name each in the "guix package -i *" declaration
> to install, but if there was some metapackage, like buildessential (I come
> from Debian).

Right.  There’s one such meta-package, which is ‘gcc-toolchain’.
Perhaps more are needed.

OTOH, ‘guix environment’ can really do “meta-packages on demand”, so to
speak, so I wonder whether things like ‘build-essential’ make so much
sense in that context.

WDYT?

Thanks,
Ludo’.

Re: [PATCH] gnu: Add most debian patches to nvi.

[Ludovic Courtès]

Marek Benc  skribis:

> On 01/22/2015 05:05 PM, Ludovic Courtès wrote:
>> Marek Benc  skribis:
>>
>>> On 01/21/2015 04:03 PM, Ludovic Courtès wrote:
 I agree with Andreas: there are 20 patches, and that level of patching
 goes beyond our mission, IMO.

 I would rather leave it as is if it’s usable, or remove it, or point to
 a new upstream if there is one.
>>> In that case, I think at least two of the patches should be allowed to
>>> get in, the ones about the Berkely db library, since it's spewing out
>>> errors without them. (but the editor itself works)
>> You mean there are run-time errors without nvi-db4.patch?  What kind?
>
> I mistyped, it's only warnings, the editor seems to work okay. In
> particular,
> the warnings look like this:
>
> BDB0635 DB_CREATE must be specified to create databases.
> BDB0511 page sizes must be a power-of-2

OK.

>> The second patch is about page sizes apparently.  Do you know what the
>> deal is?
>
> nvi seems to store the data of files in Berkely DB database objects,
> and when
> setting up a database for a file, it determines a page size to fit the file
> into 15 pages or less, using a page size of at least 1K and at most 10K.
>
> The way it does the calculation, I quote:
>   psize = ((sb.st_size / 15) + 1023) / 1024;
>   if (psize > 10)
> psize = 10;
>   if (psize == 0)
> psize = 1;
>   psize *= 1024;
>
> makes psize hold a variable that's a multiple of 1024, but not necessarilya
> power of two (for example, 3072). This didn't use to be a problem
> before, but
> nowadays, BDB requires pages to be powers of two, and so the modified
> code, I

Oh, OK.

Could you quote the warnings and fold these explanations at the top of
the patches?  And then we’ll be all set.

Thanks for the explanation,
Ludo’.

Re: A couple roadblocks using guix

[Thompson, David]

On Thu, Jan 22, 2015 at 3:49 PM, Ludovic Courtès  wrote:
> Omar Radwan  skribis:
>
>> The last thing is that I have not found any meta-packages, which allow me
>> to install multiple packages that are needed for the same program, like for
>> instance, there is gcc, make, cmake, and all the bread-and-butter build
>> programs, but you have to name each in the "guix package -i *" declaration
>> to install, but if there was some metapackage, like buildessential (I come
>> from Debian).
>
> Right.  There’s one such meta-package, which is ‘gcc-toolchain’.
> Perhaps more are needed.
>
> OTOH, ‘guix environment’ can really do “meta-packages on demand”, so to
> speak, so I wonder whether things like ‘build-essential’ make so much
> sense in that context.

However, because 'guix environment' doesn't manipulate a user's
profile, I think that it would still be very useful to create these
meta-packages.  For this specific case, yes, 'guix environment' can be
used to fetch the necessary GCC toolchain to develop something, but
what about non-development packages?  It would be very convenient to
be able to run 'guix package -i xfce-desktop', for example.

My 2 cents.

- Dave

Re: A couple roadblocks using guix

[Omar Radwan]

>I think this is the same problem as >http://bugs.gnu.org/19119>
bugs.gnu.org /19119 >,
which
>was fixed a while back.

>This suggests that it’s high time for a >release.  ;-)

I don't think there where any big features added that are worthy of a 0.9
release yet. Its just a couple package updates, bug fixes, and a couple
feature additions. I think a 0.8.1 would be adequate.

Re: Circular dependencies in python modules

[Cyril Roelandt]

On 01/22/2015 08:37 PM, Andreas Enge wrote:
> On Thu, Jan 22, 2015 at 09:50:35AM +0100, Federico Beffa wrote:
>> It appears that SPARQLWrapper is not really needed and you can disable
>> dependency guessing with '--no-guessing-deps'. See
>> https://sources.debian.net/src/rdflib/4.1.2-3/debian/rules/
> 
> Thanks for the suggestion! This seems to be some debian specific macro in
> their package recipes; I tried it as a configure flag, and setup.py chokes
> on it.
> 
> However, there are these lines in setup.py:
> kwargs['install_requires'] = [
> 'isodate',
> 'pyparsing', 'SPARQLWrapper']
> explicitly in the "else" branch of
>   if sys.version_info[0] >= 3:
> So this would explain why there was no problem with python-3.
> 

In Python 3, it is not in "install_requires", but it is in "requires":

https://github.com/RDFLib/rdflib/blob/master/setup.py#L43

Not sure exactly what the difference is, though.


Cyril.

Re: [PATCH] gnu: Add most debian patches to nvi.

[Marek Benc]

On 01/22/2015 09:52 PM, Ludovic Courtès wrote:

Marek Benc  skribis:


On 01/22/2015 05:05 PM, Ludovic Courtès wrote:

Marek Benc  skribis:


On 01/21/2015 04:03 PM, Ludovic Courtès wrote:

I agree with Andreas: there are 20 patches, and that level of patching
goes beyond our mission, IMO.

I would rather leave it as is if it’s usable, or remove it, or point to
a new upstream if there is one.

In that case, I think at least two of the patches should be allowed to
get in, the ones about the Berkely db library, since it's spewing out
errors without them. (but the editor itself works)

You mean there are run-time errors without nvi-db4.patch?  What kind?

I mistyped, it's only warnings, the editor seems to work okay. In
particular,
the warnings look like this:

BDB0635 DB_CREATE must be specified to create databases.
BDB0511 page sizes must be a power-of-2

OK.


The second patch is about page sizes apparently.  Do you know what the
deal is?

nvi seems to store the data of files in Berkely DB database objects,
and when
setting up a database for a file, it determines a page size to fit the file
into 15 pages or less, using a page size of at least 1K and at most 10K.

The way it does the calculation, I quote:
   psize = ((sb.st_size / 15) + 1023) / 1024;
   if (psize > 10)
 psize = 10;
   if (psize == 0)
 psize = 1;
   psize *= 1024;

makes psize hold a variable that's a multiple of 1024, but not necessarilya
power of two (for example, 3072). This didn't use to be a problem
before, but
nowadays, BDB requires pages to be powers of two, and so the modified
code, I

Oh, OK.

Could you quote the warnings and fold these explanations at the top of
the patches?  And then we’ll be all set.


Sure thing, they're in now.



Thanks for the explanation,
Ludo’.


No problem, and thanks,
Marek.
>From e101c036615bb22b05744789fbc0f5ddb7f71413 Mon Sep 17 00:00:00 2001
From: Marek Benc 
Date: Thu, 22 Jan 2015 23:08:16 +0100
Subject: [PATCH] gnu: nvi: Fix Berkely DB compatability issues.

* gnu/packages/patches/nvi-db4.patch: New file.
* gnu/packages/patches/nvi-dbpagesize-binpower.patch: New file.
* gnu/packages/nvi.scm (nvi): Make use of them.
* gnu-system.am (dist_patch_DATA): Add them.

---
 gnu-system.am  |  4 ++-
 gnu/packages/nvi.scm   |  4 ++-
 gnu/packages/patches/nvi-db4.patch | 35 ++
 gnu/packages/patches/nvi-dbpagesize-binpower.patch | 35 ++
 4 files changed, 76 insertions(+), 2 deletions(-)
 create mode 100644 gnu/packages/patches/nvi-db4.patch
 create mode 100644 gnu/packages/patches/nvi-dbpagesize-binpower.patch

diff --git a/gnu-system.am b/gnu-system.am
index d1fc59a..5d773ca 100644
--- a/gnu-system.am
+++ b/gnu-system.am
@@ -443,7 +443,9 @@ dist_patch_DATA =		\
   gnu/packages/patches/net-tools-bitrot.patch			\
   gnu/packages/patches/ninja-tests.patch			\
   gnu/packages/patches/nss-pkgconfig.patch			\
-  gnu/packages/patches/nvi-assume-preserve-path.patch   \
+  gnu/packages/patches/nvi-assume-preserve-path.patch		\
+  gnu/packages/patches/nvi-dbpagesize-binpower.patch		\
+  gnu/packages/patches/nvi-db4.patch\
   gnu/packages/patches/orpheus-cast-errors-and-includes.patch	\
   gnu/packages/patches/ots-no-include-missing-file.patch	\
   gnu/packages/patches/patchelf-page-size.patch			\
diff --git a/gnu/packages/nvi.scm b/gnu/packages/nvi.scm
index 8fd736d..128715f 100644
--- a/gnu/packages/nvi.scm
+++ b/gnu/packages/nvi.scm
@@ -37,7 +37,9 @@
  ".tar.bz2"))
 (sha256
   (base32 "0nbbs1inyrqds0ywn3ln5slv54v5zraq7lszkg8nsavv4kivhh9l"))
-(patches (list (search-patch "nvi-assume-preserve-path.patch")))
+(patches (list (search-patch "nvi-assume-preserve-path.patch")
+   (search-patch "nvi-dbpagesize-binpower.patch")
+   (search-patch "nvi-db4.patch")))
 (snippet
   ;; Create a wrapper for the configure script, make it executable.
   '(let ((conf-wrap (open-output-file "configure")))
diff --git a/gnu/packages/patches/nvi-db4.patch b/gnu/packages/patches/nvi-db4.patch
new file mode 100644
index 000..03b736c
--- /dev/null
+++ b/gnu/packages/patches/nvi-db4.patch
@@ -0,0 +1,35 @@
+This patch originates from the Debian project, see https://www.debian.org/
+
+03db4.dpatch  by  
+
+
+libdb4 compatibility adjustments.
+
+In particular, this patch adds extra file permission checking and passes the
+DB_CREATE flag to the first invocation of db_open on the file's database
+structure, which rids us of the following message:
+
+  BDB0635 DB_CREATE must be specified to create databases.
+
+--- nvi-1.81.6.orig/common/msg.c	2009-02-26 14:26:58.350336128 +0100
 nvi-1.81.6/common/msg.c	2009-02-26 14:29:05.235335829 +0100
+@@ -724,9 +724,18 @@
+ 		p = buf;
+ 	} else
+ 		p = file;
++	if (access(p, F_OK) != 0) {
++		if (first) {
++			first = 0;
++			return (1);
++		}
++

Re: ROADMAP in git tree needs to be updated.

[Ludovic Courtès]

l...@gnu.org (Ludovic Courtès) skribis:

> In place of a road map, here’s my take at a relatively detailed list of
> things that I think we should have for 1.0 to satisfy a range of use
> cases, and things that can happen later:

Committed.

Ludo’.

guix.el works out of the box

[Ludovic Courtès]

With commit 0a05176, which creates a simple site-start.el and sets
EMACSLOADPATH appropriately, users are able to fire Emacs and from there
guix.el works out-of-the-box, using the .el and .scm files available in
/run/current-system/profile.

Comments welcome!

Ludo’.

Re: A couple roadblocks using guix

[Omar Radwan]

>I think this is the same problem as , which
>was fixed a while back.

Is there any way to fix this problem for the time being without having to
wait for a release?

On Thu, Jan 22, 2015 at 12:49 PM, Ludovic Courtès  wrote:

> Omar Radwan  skribis:
>
> > One of the things is that during system initialization, the (gnu system
> > networking) module returns an error which reads "guix system: error:
> failed
> > to load operating system file '/mnt/etc/config.scm': (misc-error #f "~A
> ~S"
> > ("no code for module" (gnu system networking)) #f)
> > and that kind of makes things really annoying to manually restart
> > networking every reboot.
>
> There’s a typo in the 0.8 manual (which was fixed in fa1e31b8): it
> should read (gnu services networking).
>
> > Another thing is that is sort of a problem is during a windowmaker X
> > session, all my graphical applications(xterm, icecat, etc), are not
> listed
> > in the application menu and are not runnable
>
> I think this is the same problem as , which
> was fixed a while back.
>
> This suggests that it’s high time for a release.  ;-)
>
> > The last thing is that I have not found any meta-packages, which allow me
> > to install multiple packages that are needed for the same program, like
> for
> > instance, there is gcc, make, cmake, and all the bread-and-butter build
> > programs, but you have to name each in the "guix package -i *"
> declaration
> > to install, but if there was some metapackage, like buildessential (I
> come
> > from Debian).
>
> Right.  There’s one such meta-package, which is ‘gcc-toolchain’.
> Perhaps more are needed.
>
> OTOH, ‘guix environment’ can really do “meta-packages on demand”, so to
> speak, so I wonder whether things like ‘build-essential’ make so much
> sense in that context.
>
> WDYT?
>
> Thanks,
> Ludo’.
>