Re: apt-get install wordperfect?

1999-05-27 Thread James Mastros
On Wed, May 26, 1999 at 08:01:27PM -0500, Mark Mealman wrote:
> Is it possible for commercial software to make it into the Debian
> archives(presumably in non-free)?
[...]

I'd tend to agree with the feeling in the above (apt-able archives of
non-(DFSG)-free but (beer)-free are a Good Thing), but don't like the idea
of Debian doing it -- it just dosn't seem like your (Caveat empator: I am
not a debian-developer (yet)) place, and it almost certianly wasn't the idea
of the people donating server space/bandwidth to be hosting commercial
software.  Perhaps all we need to do is write a nice easy
commercial-software-packaging HOWTO, complete from writing a workable
package (which dosn't have to comply with debain policy, making it /really/
easy) to creating a nice place to put them with a Packages.gz.

Make it trivial for the commercial people to let their customers use apt's
slickness, and perhaps they will... and we certianly wouldn't mind.

-=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Re: Source-depends?

1999-05-23 Thread James Mastros
On Sat, May 22, 1999 at 04:38:21AM +0930, Ron wrote:
> > Well, it sounds like you repeated what about a dozen people have already
> > said. The concern is an automated way to generate the depends. 
Umm, any purticular reason to that compile-depends must be autogenerated --
why can't they be done manualy by the packager?  (I realize that this is less
then ideal -- but having this in optionaly, manualy seems better then not
having it at all.)

> > The
> > autobuilders already use a semi-working type of this, but it isn't perfect
> > and makes assumptions that can't always be assumed.
Mind describing the method the autobuilders use (and where I can find
approprate source)?

> > I have already made a patch for dpkg-* programs to use source deps in a
> > control field, that's not the problem though.
Any reason not to put it in your next upload?

> As this has probably been thrashed about innumerable times before, yet we
> all agree that having source dependancies would be valuable, could someone
> summarise (or provide a pointer to) the problems that have been identified.
> 
> I see two situations up front:
>  - a need to describe the tools needed to build a package
> (eg. gcc, bison, flex, etc..)
>  - and a need to describe the other source packages or librarys required
> to build a working binary.
Why do these need to be treated differently?

> as well as a way to auto-detect these dependancies, what else is required?
I've read through all of the archives that seemed appropriate, and found the
following problems that had been brought up:
1. Automatic generation of the field(s)
2. Required/Recomened/Sugueted defs (for ex, what to do with tex tools and
   such that are required for only minor peices of the packages.)
3. Requiring bin-packages vs. requiring build trees from other package's
   source

In general, on prior discussions, there was no conclusion on 1 and 2, but on
3 it was suguested that requiring build trees from another package was
considered buggy.

A soultion to number 1 that was tossed around included using libtricks to
get a list of files accessed, and is therefor (IIRC) obselete.  (And in any
case is prohibitively slow.)

-=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Source-depends?

1999-05-21 Thread James Mastros
I've noticed that many people on this list are, of late, talking about
source-depends, but not doing anything about standardizing it... thus, a
psudo-proposal for you guys (psudo because I can't formaly propose anything;
I'm not yet a developer):
A new control-file field be added, Compilation-Depends, which lists all
packages required to successfully compile a working version of the package
without major changes to the source (IE editing a Makefile to change a
clearly documented option or change a ./configure option is OK, but having to 
rip things out of the source
is not).  Another be added, Compilation-Suguests, that lists all packages
needed to compile the package to get functionality equivlent to the
cannonical binary-package.  (For example, libgnomeui-dev would be a
Compilation-Suguests, but not a Compilation-Depends, for packages that are
optionaly gnomeified.)  In the case of multi-binary packages, the "major"
binary should be compilable with only the Compilation-Depends.  (For
example, in the case of the source-package enlightenment, the binary-package
enlightenment would be the primary).  (I would simply make the least
required for any binary-package be Depends, but that would be obviously
wrong for -dev and -doc packages.)  The Compilation-{Depends, Suguests}
should be propagated to the .dsc files, but not the .deb files; they aren't
useful in binary-packages.

Once the fields are standardized, the auto-builders should be able to use
them fairly easily, and lintian could allow them trivialy.  (I would tend to
say that apt-get source shouldn't use them, as getting the source as a
reference, without wanting to compile it, is probably fairly common.
Anyway, that's a call for apt's maintaner.)

So how does this all sound to you guys?

-=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Bug#38057: general: libgtop0: Depends: libglib1.1.13 (>= 1.1.13-1) but it is not installable

1999-05-20 Thread James Mastros
On Thu, May 20, 1999 at 04:47:13PM -0400, Sean Chuplis wrote:
> Sorry, but the following packages have unmet dependencies:
>   libgtop0: Depends: libglib1.1.13 (>= 1.1.13-1) but it is not
> installable
libgtop0 should be removed from the archive; it is obselete and replaced by
libgtop1.  gnome-utils 0.99.3-1 depends on it -- but gnome-utils 0.99.3-1 is
also obselete, but I cannot find a replacement, even though I have the
replacement installed!  Is it still stuck in Incoming?

    -=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Re: Two sets of packages for slink and potato. How to version?

1999-05-19 Thread James Mastros
On Tue, May 18, 1999 at 03:45:23PM +0200, Stephane Bortzmeyer wrote:
> The problem is the versioning. How to choose the version numbers in the two 
> sets so that users will automatically get the potato package when they will 
> choose to replace 'stable' by 'unstable' (or when potato will become stable).
> 
> I've read 
> <http://www.debian.org/doc/packaging-manuals/packaging.html/ch-versio
> ns.html>. Should I choose an epoch of 1 for all the potato packages?
In general, epochs are Considered Dangerous (for reasons I don't really
understand -- but I assume that they are good).  I'd suguest subtracting .01
from the debian-version and concatinating '.slink' to the end (somthing like
1.0-0.99.slink); that seems to be standard pratice.

-=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Re: Hints about future improvements

1999-05-19 Thread James Mastros
On Wed, May 19, 1999 at 02:51:20AM +0200, Gabor Fleischer wrote:
> I was thinking about this, and there's one thing which is not the
> best it could be, I think: Let's see libc, and the packages that
> are compiled from the same source, for example locales.
[...]
> 
> There could be a value in the control file like: Last-changed-version or
> something similar. apt/dselect could decide from this wether it
> needs to download this package or not.
[...]
Or, we could try to depricate giving the source-package and all of it's
associated binary-packages the same version number, even when there are no
changes -- more work for the maintainer, without a dought (more places to
change the version-number, needing to check in which bin-package the version
number should increase, and what _versioned_ interdependincies the packages
should have... but it would significantly decrease the bulk of packages that
need to be upgraed (glibc-doc and libc6-dev especially).

-=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Re: Compiling wordinspect for potato

1999-05-17 Thread James Mastros
On Mon, May 17, 1999 at 05:02:30PM -0400, Bob Hilliard wrote:
>  I installed libglib1.2_1.2.3-1.deb, libglib1.2-dev_1.2.3-1.deb,
> libgtk1.2_1.2.3-1.deb, and libgtk1.2-dev_1.2.3-1.deb, and built the
> package.  It compiled cleanly, and runs as it did in slink, except
> that every time it pops up a window it displays the following message:
> 
> > Gtk-WARNING **: gtk_scrolled_window_add(): cannot add non scrollable widget
> > use gtk_scrolled_window_add_with_viewport() instead
> 
>  Would another libgtk be a better choice to replace libgtk1.1,
> and possibly get away from this warning message, or will the source
> have to tweaked to avoid this?

You'll have to make some changes to the source; they aren't too difficult --
I don't remember specifics, though.  I'd stick with gtk (et al) 1.2; things
in potato depending on obselete libraries is a Bad Thing when at all
avoidable.

BTW, debian-gtk-gnome@lists.debian.org is the cannonical place to ask
debian-related gtk questions, and it's low-volume now that the Great GNOME
Copy is done.  (CCed/Reply-toed there.)

-=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Re: Splitting debian-devel-changes to separate lists

1999-05-14 Thread James Mastros
On Thu, May 13, 1999 at 03:37:32PM -0700, Joseph Carter wrote:
> On Thu, May 13, 1999 at 12:59:12PM +0200, Wichert Akkerman wrote:
> > :0
> > * ^Subject:.*\((alpha|arm|powerpc|m68k|sparc)\)
> > /dev/null
> 
> A slight mod:
> 
> :0
> * ^Subject:.*\((alpha|arm|powerpc|m68k|sparc)\)
> * !^Subject:.*source
> /dev/null

The second regex would seem to be unnecessary; the '|'s will allow only one,
and exactly one, of the choices, and the parens are right next to the or-list.

I'm using this:
:0
* ^Subject:[[:space:]]*Uploaded 
[1-90.-+][[:space:]]\((alpha|arm|powerpc|m68k|sparc|hurd-i386)\)
/dev/null

This is more specific, and therefor less likely to catch other mail.  (also,
I added hurd-i386 to the list -- forgot that one.)

-=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo



Slink compiles [was: Re: Release Plans (1999-05-10)]

1999-05-12 Thread James Mastros
On Wed, May 12, 1999 at 02:40:33PM -0700, Darren O. Benham wrote:
> Propose it on -Policy and be sure the cc' the ftpmasters.  An issue to
> consider is manpower.  Since most of the maintainers will up to Potato,
> who'll compile these new packages/updates for slink?
Can't autobuilders do it (with a source-only upload)?

    -=- James Mastros
-- 
First they came for the fourth amendment, but I said nothing because I
wasn't a drug dealer. Then they came for the sixth amendment, but I kept
quiet because I wasn't guilty. Finally they came for the first amendment,
and by then it was too late to say anything at all." 
-=- Nancy Lebowitz
cat /dev/urandom|james --insane=yes > http://www.rtweb.net/theorb/
ICQ: 1293899   AIM: theorbtwo  YPager: theorbtwo