Re: PLEASE TEST: pkgtools changes

Dixitur illum bsiegert@... scribere...

>Apply the patch (see below) and test it. It would also be nice if someone
>could see through the patch and check for mistakes. FWIW, it builds with

 <at>  <at>  -38,6 +38,7  <at>  <at> 
        char    *cp;
        char    name[FILENAME_MAX];
        char    buf[LegibleChecksumLen];
+       int len;

        for (p = pkg->head ; p ; p = p->next) {
                switch (p->type) {

-> indenting

+                       break;
+               case PLIST_INFO:

-> I'd rather put the PLIST_INFO case _before_ the PLIST_FILE case
   and have it fall through, i.e. no break; statement.
   This gets rid of the "there is no MD5 for info files" problem
   you told me about, and reduces duplicate code.

Two general problems I _think_ might still be there:

* in OpenBSD,  <at> info foo.info scans for foo.info-1, foo.info-2 etc.
  and also adds them
  (we prefer --no-split anyway, but...)
* the  <at> dirrm list must be reversely sorted, you might want to use

Acrobat reader, Linuxulator

Hi all!

With -current of today (HERC), as in
tg <at> odem:/home/tg $ ident /sys/compat/linux/linux_sysent.c                                                       
/sys/compat/linux/linux_sysent.c:
     $MirBSD: src/sys/compat/linux/linux_sysent.c,v 1.11 2004/08/02 18:27:31 tg Exp $

I've fixed the linuxulator well enough that it runs most simple stuff.

Acroread doesn't still work: try to press ctrl-k; instead of a
preferences window opening, you get a 1px window of shit, and
nothing works any more (beeps). Same if I run it out of a bash
linux binary.

It works well if run from kdump.

WTF???

I have no idea how to fix, no, actually _debug_ this. Can please
someone help with that?

bye,
//mirabile

Re: PLEASE TEST: pkgtools changes

On Mon, 2 Aug 2004, Thorsten Glaser wrote:

> -> indenting

*sigh*

> -> I'd rather put the PLIST_INFO case _before_ the PLIST_FILE case
>    and have it fall through, i.e. no break; statement.
>    This gets rid of the "there is no MD5 for info files" problem
>    you told me about, and reduces duplicate code.

Great idea.

> * the  <at> dirrm list must be reversely sorted, you might want to use
>   a LIFO buffer for them (stack based) -> but then, what do you
>   do about the  <at> dirrms which are already in?

How about the follwing idea: I add the  <at> dirrms before the first other 
 <at> dirrm. This way,

foo/
foo/bar/
 <at> dirrm baz

becomes

[...]
 <at> dirrm foo/bar
 <at> dirrm foo
 <at> dirrm baz

Re: PLEASE TEST: pkgtools changes

Dixitur illum bsiegert@... scribere...

>How about the follwing idea: I add the  <at> dirrms before the first other
> <at> dirrm. This way,
[...]
>We already do. It is in a doubly linked list.

I'd rather reorder that list.

perls sort function looks like broken sometimes, too - for
example, it has tried to fubar the mlterm PFRAG.shared when
I fixed that port, putting the libkik/libmkf below the
plugins.

OTOH it's you who is coding, so I don't know if you want to
put in that much effort.

//Thorsten
--

-- 
Currently blocking eMail from the following domains: bigpond.com, biz, gmx.de,
gmx.net, hotmail.com, info, jumpy.it, libero.it, name, netscape.net,
postino.it, simplesnet.pt, spymac.com, tatanova.com, tiscali.co.uk,
tiscali.cz, tiscali.de, tiscali.it, voila.fr, yahoo.co.uk, yahoo.com.

Re: Acrobat reader, Linuxulator

On Mon, 2 Aug 2004 18:52:27 GMT, Thorsten Glaser wrote
> I've fixed the linuxulator well enough that it runs most 
> simple stuff.

That's good to hear.

> Acroread doesn't still work: try to press ctrl-k; instead of a
> preferences window opening, you get a 1px window of shit, and
> nothing works any more (beeps). Same if I run it out of a bash
> linux binary.
> 
> It works well if run from kdump.
> 
> WTF???
> 
> I have no idea how to fix, no, actually _debug_ this. Can 
> please someone help with that?

I might have a chance to take a look at this in a couple of days, though
if it works while it's system calls are being traced it'll be a hard
problem to narrow down.  I've got a little program that uses ptrace() to
log certain system calls that I should be able to expand for this, maybe
it won't cause the problem to disappear like kdump does.  But before I
look at that problem I've got to deal with a little networking problem
I'm having.

--

-- 
Stephen Paskaluk
sap@...

Re: Acrobat reader, Linuxulator

Dixitur illum sap@... scribere...

>I might have a chance to take a look at this in a couple of days, though
>if it works while it's system calls are being traced it'll be a hard
>problem to narrow down.  I've got a little program that uses ptrace() to
>log certain system calls that I should be able to expand for this, maybe
>it won't cause the problem to disappear like kdump does.  But before I
>look at that problem I've got to deal with a little networking problem
>I'm having.

Thanks and good luck. I'll be pretty offline for another few days
for family reasons, so I can only peek into mail and IRC once in
a while, but not hack much. I'll try to, though.

//Thorsten
--

-- 
Currently blocking eMail from the following domains: bigpond.com, biz, gmx.de,
gmx.net, hotmail.com, info, jumpy.it, libero.it, name, netscape.net,
postino.it, simplesnet.pt, spymac.com, tatanova.com, tiscali.co.uk,
tiscali.cz, tiscali.de, tiscali.it, voila.fr, yahoo.co.uk, yahoo.com.

Fwd: CVS: cvs.openbsd.org: src

It's enough.

I vote for:
- no changes to our pkgtools to support these keywords
- one, last, global import of OpenBSD ports
  (probably this month)
- total regeneration of all PLISTs

Benny, sorry for all the effort you put into the pkgtools,
but maybe they'll benefit from it anyway.

(Are we going to use dir/ ? It's the _only_ one of the new
 changes which looks quite useful to me.)

//Thorsten
--

-- 
Currently blocking eMail from the following domains: bigpond.com, biz, gmx.de,
gmx.net, hotmail.com, info, jumpy.it, libero.it, name, netscape.net,
postino.it, simplesnet.pt, spymac.com, tatanova.com, tiscali.co.uk,
tiscali.cz, tiscali.de, tiscali.it, voila.fr, yahoo.co.uk, yahoo.com.

---------- Forwarded message ----------
From: Marc Espie <espie@...>
Message-ID: <200408031229.i73CTj05024120@...>
Newsgroups: gmane.os.openbsd.cvs
Date: Tue, 3 Aug 2004 06:29:45 -0600 (MDT)
Subject: CVS: cvs.openbsd.org: src

CVSROOT:	/cvs
Module name:	src

Future of MirPorts (was: Re: Fwd: CVS: cvs.openbsd.org: src)

On Wed, 4 Aug 2004, Thorsten Glaser wrote:

> It's enough.
> 
> I vote for:
> - no changes to our pkgtools to support these keywords
> - one, last, global import of OpenBSD ports
>   (probably this month)

I tend to disagree. I am very reluctant to give up the regular imports. 
The reason for this has been discussed before: It takes a lot more than us 
four or so developers to maintain an _entire_ ports tree. Just see how 
long we have been having the MirBSD uname -- and still there are lots of 
unfixed ports.

Somehow I would also be willing to use the new pkgtools. If you look 
close, the code is easy to read and to maintain -- even with my very 
limited knowledge of Perl.

I also like the  <at> sample keyword. 90% of the INSTALL scripts deal with 
config file installation. Thus, it is an obvious thing to add to the 
package tools. Debian has a vaguely similar mechanism, which I found very 
useful. If we keep the old pkgtools, I would implement this.

--Benny.

Re: Future of MirPorts

Dixitur illum bsiegert@... scribere...

>I tend to disagree. I am very reluctant to give up the regular imports.
>The reason for this has been discussed before: It takes a lot more than us
>four or so developers to maintain an _entire_ ports tree. Just see how

Each time I read openbsd ports-changes <at> , in my inner eye I see that
the next merge will take up even more time.

Also, I *do not* like the direction the OpenBSD people go.

>Somehow I would also be willing to use the new pkgtools. If you look

I'd rather write different new pkgtools in a real language.

Also, who of us two has been complaining that Espie changes things
around without announcing that anything has changed, even weeks
after that?

------------------------------------------------------------
| I want to cut our dependency on OpenBSD at some point in |
| the future. Early for ports, late for source.            |
------------------------------------------------------------

>I also like the  <at> sample keyword. 90% of the INSTALL scripts deal with
>config file installation. Thus, it is an obvious thing to add to the
>package tools. Debian has a vaguely similar mechanism, which I found very
>useful. If we keep the old pkgtools, I would implement this.

Your argumentation is right.

Regarding mirbsd.bsdadvocacy.org/?features