Re: Building a build environment

2008-07-30 Thread Eric Pozharski
On Wed, Jul 30, 2008 at 10:04:35AM +1000, Ben Finney wrote:
> Richard Hurt <[EMAIL PROTECTED]> writes:

*SKIP*
> > Also, the wide array of tools available to me is pretty
> > overwhelming. pBuilder, cowbuilder, cowdancer, quilt, etc... The
> > list is quite long and I'm not sure which one is old, new, mature,
> > good or bad. Does anyone have any advice for a newbie packager that
> > just wants a simple setup that works? What do you use?
> 
> The Developer's Reference and New Maintainer's Guide should be your
> first port of call. Read them thoroughly and become familiar with
> them; they have much advice on tools and procedures.

[To: OP] I would like to stress on that point.  You *must* be familiar
with Debian internals, with ways Debian builds from source, with Debian
BTS.  When something will go wrong, you'll be on your own.

-- 
Torvalds' goal for Linux is very simple: World Domination


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Please clarify: to bugreport or not bugreport

2008-11-01 Thread Eric Pozharski
Just for the record;  I routinely build packages from debian source
(that's my way to testing).  Each build starts from minimal environment
(dpkg, apt, build-essential, their dependencies);  then Build-Depends is
satisfied and build goes on.  I suppose that either environment on
build-machines where binaries for Debian are built is not that minimal
or Build-Depends is satisfied not that blindly.

Hence the problem;  Sometimes I FTBFS some package -- because
Build-Depends either misses some package or lacks explicitly set
relation.

FTBFS itself doesn't make problem for me (each time I'd managed to
figure out a block and workaround it manually).  I still can't decide --
should (would? could?) I bugreport this?

Look, I admit that I build unusual way.  Maybe it's not the problem?
(I've solved that twice today, that's why I post.)

-- 
Torvalds' goal for Linux is very simple: World Domination


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Please clarify: to bugreport or not bugreport

2008-11-12 Thread Eric Pozharski
On Tue, Nov 11, 2008 at 11:50:27AM +0100, Cyril Brulebois wrote:
> Paul Wise <[EMAIL PROTECTED]> (02/11/2008):
> > If the build-deps are satisfied and build-essential is installed, the
> > package should build, so yeah, file bugs at severity serious please.
> 
> For reference:
> 
>  Do folks think #505040 is RC or should at least be
>fixed for lenny?
>  KiBi: nope, the package in etch builds in etch, the one
>in lenny builds in lenny, so it's not considered RC, though
>fixing it won't harm

I'd been surprised that I should be C.  Should it be
C?  (just for the record, I'm going to bugreport B.)

-- 
Torvalds' goal for Linux is very simple: World Domination


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: Using pbuilder to test packages with gcc-snapshot

2008-11-14 Thread Eric Pozharski
On Thu, Nov 13, 2008 at 08:19:38PM -0600, Kumar Appaiah wrote:
> Dear Debian Mentors,

*SKIP*

> 1. Is there a way to set an arbitrary environment variable while
>running pbuilder (in my case, LD_LIBRARY_PATH).

Try adding C in
pbuilderrc (whatever is effective in your installation).  If my
understanding of pbuilder isn't outdated the effective pbuilderrc is
sourced in chroot context.

*CUT*


-- 
Torvalds' goal for Linux is very simple: World Domination


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]



Re: pdebuild, not removing build environment

2009-04-10 Thread Eric Pozharski
On Fri, Apr 10, 2009 at 12:36:50AM +0800, Chow Loong Jin wrote:
> On Thu, 2009-04-09 at 18:21 +0200, أحمد المحمودي wrote:
> > On Thu, Apr 09, 2009 at 04:51:35PM +0200, Gudjon I. Gudjonsson wrote:
> > > Sorry if I have missed something obvious but, in case of error during 
> > > build. How can I keep the pbuilder environment when I use pdebuild. I 
> > > don't 
> > > want pdebuild to remove the chroot environment after the error occurs.
> > ---end quoted text---
> > 
> >   Well, the only method that I know is do a pbuilder login, then I build 
> >   the package (after manually apt-get'ing its Build-Deps), so:
> > 
> >   $ pbuilder login
> >   # apt-get install 
> >   # dpkg-source -x .dsc
> >   # cd 
> >   # ./debian/rules build (or whatever)
> 
> A hook would save you all that trouble. I have a hook called C00Bash
> which looks like this:
> 8<=
> #!/bin/sh
> exec bash
> >8=
> 
> After a build (only if there's a failure), it'll drop to a Bash prompt
> within the chroot, whereby you can head to /tmp/buildd/ and examine why
> things went the way they did.

(just my 2cent) Someone interested could develop a hook that after
looking for parent processes (it should go far enough) just would kill
pbuilder.  This must be SIGKILL; otherwise the signal could be trapped.

p.s I'm not interested -- so I'm not that someone.



-- 
Torvalds' goal for Linux is very simple: World Domination
Stallman's goal for GNU is even simpler: Freedom


-- 
To UNSUBSCRIBE, email to debian-mentors-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org



Re: pdebuild, not removing build environment

2009-04-11 Thread Eric Pozharski
On Fri, Apr 10, 2009 at 10:14:07AM +, Chow Loong Jin wrote:
> On Fri, 2009-04-10 at 11:39 +0300, Eric Pozharski wrote:
> > On Fri, Apr 10, 2009 at 12:36:50AM +0800, Chow Loong Jin wrote:
> > > On Thu, 2009-04-09 at 18:21 +0200, أحمد المحمودي wrote:
> > > > On Thu, Apr 09, 2009 at 04:51:35PM +0200, Gudjon I. Gudjonsson wrote:
> > > > > Sorry if I have missed something obvious but, in case of error 
> > > > > during 
> > > > > build. How can I keep the pbuilder environment when I use pdebuild. I 
> > > > > don't 
> > > > > want pdebuild to remove the chroot environment after the error occurs.
> > > > ---end quoted text---
> > > > 
> > > >   Well, the only method that I know is do a pbuilder login, then I 
> > > > build 
> > > >   the package (after manually apt-get'ing its Build-Deps), so:
> > > > 
> > > >   $ pbuilder login
> > > >   # apt-get install 
> > > >   # dpkg-source -x .dsc
> > > >   # cd 
> > > >   # ./debian/rules build (or whatever)
> > > 
> > > A hook would save you all that trouble. I have a hook called C00Bash
> > > which looks like this:
> > > 8<=
> > > #!/bin/sh
> > > exec bash
> > > >8=
> > > 
> > > After a build (only if there's a failure), it'll drop to a Bash prompt
> > > within the chroot, whereby you can head to /tmp/buildd/ and examine why
> > > things went the way they did.
> > 
> > (just my 2cent) Someone interested could develop a hook that after
> > looking for parent processes (it should go far enough) just would kill
> > pbuilder.  This must be SIGKILL; otherwise the signal could be trapped.
> > 
> > p.s I'm not interested -- so I'm not that someone.
> > 
> Just replace "exec bash" with "killall -9 pbuilder". But seriously it's
> pointless. If you keep the Bash prompt from that hook I posted earlier
> open, you can poke around inside the chroot, and if you wish, you could
> do the killing of pbuilder yourself, as it'll wait for you to close that
> prompt.

Exactly.  But OP wanted to keep tree (for unknown reason).



-- 
Torvalds' goal for Linux is very simple: World Domination
Stallman's goal for GNU is even simpler: Freedom


-- 
To UNSUBSCRIBE, email to debian-mentors-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org