Re: Understanding versioning.
On Tue, 08 Nov 2011 15:56:02 -0500, Rick Thomas wrote: > On Nov 8, 2011, at 2:07 AM, Sthu Deus wrote: > >> it seems to me to be weird having those "epoches" > > If all software developers were "well behaved" and they all co- operated > in their versioning, it would be weird to have "epochs". All versions, > from all sources, would be monotonically increasing as a function of > time. > > But, as we know, software developers are an independent bunch of > cowboys. Sometimes, for reasons that seem good and reasonable to the > individual upstream developer involved, software version time seems to > run backwards. "Epochs" are there so the rest of us (who aren't > upstream developers) can pretend it didn't do that -- when the > developers glitch their versions, we just move into a different epoch > and go on about our lives. It might be less confusing for the uninitiated though, if the term used were "incarnation" rather than "epoch"; at least IMHO. Especially so in the light of the other *nix use of the term epoch regarding time. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/j9c9cb$3jh$1...@dough.gmane.org
Re: Understanding versioning.
On Nov 8, 2011, at 2:07 AM, Sthu Deus wrote: it seems to me to be weird having those "epoches" If all software developers were "well behaved" and they all co- operated in their versioning, it would be weird to have "epochs". All versions, from all sources, would be monotonically increasing as a function of time. But, as we know, software developers are an independent bunch of cowboys. Sometimes, for reasons that seem good and reasonable to the individual upstream developer involved, software version time seems to run backwards. "Epochs" are there so the rest of us (who aren't upstream developers) can pretend it didn't do that -- when the developers glitch their versions, we just move into a different epoch and go on about our lives. Rick -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/edf1249b-a1d0-429f-abc5-524dd9621...@pobox.com
Re: Understanding versioning.
Thank You for Your time and answer, Dan: >First, don't think of second or fifth (per >http://dictionary.reference.com/browse/epoch) senses of "epoch"-- >the _beginning_ of some period (the meaning used re Unix time). >Think of the first or fourth senses--a _period_ of time. > >Then, think of a package version epoch as the period of time during >which a coherent (i.e., ascending) sequence of version numbers is >used. > >If such a period ends because, say because you issue a new version >number that is lower than (and therefore seems older than) an older >version number, a way to resolve the confusion is to declare that >you have switched to using a new system of version numbers. Yes, I got that. Just wanted more explanation on the matter - of course I am not an expert in software development and in industrial scale, therefore it seems to me to be weird having those "epoches" - for there is number positioning in versions like a.b.c.d - each number saying about how significant development progressed, and there is "-e.f" saying about patches in a distro being applied. But thank You, Dan, and all others that took Your time to make things clearer to me. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4eb8d55e.c820cc0a.5e23.0...@mx.google.com
Re: Understanding versioning.
Sthu Deus wrote: Thank You for Your time and answer, Camaleón: It is provided to allow mistakes in the version numbers of older versions of a package, and also a package's previous version numbering schemes, to be left behind. What does this mean? From other posts in the thread it is still not clear to me. If You can explain, please do so. Also what is epoch in this sense? I now time is measured in UNIX systems since the first system ran - it's obvious - a new great OS generation has started and till now rolls on. But what about the package epoch they use? First, don't think of second or fifth (per http://dictionary.reference.com/browse/epoch) senses of "epoch"-- the _beginning_ of some period (the meaning used re Unix time). Think of the first or fourth senses--a _period_ of time. Then, think of a package version epoch as the period of time during which a coherent (i.e., ascending) sequence of version numbers is used. If such a period ends because, say because you issue a new version number that is lower than (and therefore seems older than) an older version number, a way to resolve the confusion is to declare that you have switched to using a new system of version numbers. The package version epoch number identifies in which system of version numbers a given version number reference is written. Daniel -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4eb85a7a.8030...@kempt.net
Re: Understanding versioning.
On Mon, 07 Nov 2011 15:37:58 +0700, Sthu Deus wrote: > Thank You for Your time and answer, Camaleón: > >>It is provided to allow mistakes in the version numbers of older >>versions of a package, and also a package's previous version numbering >>schemes, to be left behind. > > What does this mean? From other posts in the thread it is still not > clear to me. If You can explain, please do so. > > Also what is epoch in this sense? I now time is measured in UNIX systems > since the first system ran - it's obvious - a new great OS generation > has started and till now rolls on. But what about the package epoch they > use? I think Henrique de Moraes already explained this quite well :-) Although I was not aware of the inners, I was taking that number as a kind of internal string to help packagers to control software versioning. Greetings, -- Camaleón -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/pan.2011.11.07.14.52...@gmail.com
Re: Understanding versioning.
Thank You for Your time and answer, Camaleón: >It is provided to allow mistakes in the version numbers of older >versions of a package, and also a package's previous version numbering >schemes, to be left behind. What does this mean? From other posts in the thread it is still not clear to me. If You can explain, please do so. Also what is epoch in this sense? I now time is measured in UNIX systems since the first system ran - it's obvious - a new great OS generation has started and till now rolls on. But what about the package epoch they use? -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4eb798f3.0611cc0a.3bf2.6...@mx.google.com
Re: Understanding versioning.
Henrique de Moraes Holschuh writes: [...] >> What does that non-sensical sounding explanation mean? Its not as if >> it is explained at the URL cited. > > It is a version override. For an epoch of "n", *any* version without an epoch > or with an epoch that is lower than "n" will be considered *older*. > > So 5:0.0.1 is NEWER than 10.2.3, or than 1:9.1.4. [...] Javier Barroso writes: [...] > Hi, > > bash-completion has a clear example of use: > See /usr/share/doc/bash-completion/changelog.Debian.gz > After bash-completion (20080705) version the next version is 1.0, so > epoch had to be used > > busybox has another example (developper simply upload an package with > incorrect version): > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=97290 > > You can have fun time investigating which packages you have installed > has epoch in their version and why they have it, running attached perl > script ( I was boring this night ...) (I'll be stealing off that for a while... hehe). Ok, that all clears things up very nicely. Thank you both for taking time to clue me in. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87lirsr91p@newsguy.com
Re: Understanding versioning.
On Sun, Nov 6, 2011 at 11:09 PM, Harry Putnam wrote:
> Camaleón writes:
>> http://www.debian.org/doc/debian-policy/ch-controlfields.html
>
> > epoch
>
>> This is a single (generally small) unsigned integer. It may be omitted,
>> in which case zero is assumed. If it is omitted then the upstream_version
>> may not contain any colons.
>>
>> It is provided to allow mistakes in the version numbers of older versions
>> of a package, and also a package's previous version numbering schemes, to
>> be left behind.
>
> What does that non-sensical sounding explanation mean? Its not as if
> it is explained at the URL cited.
>
> I'm sure its perfectly reasonable if you know enough about debian
> generally, but if not it sounds kind of off the wall.
>
> Can you show an example of how it works?
Hi,
bash-completion has a clear example of use:
See /usr/share/doc/bash-completion/changelog.Debian.gz
After bash-completion (20080705) version the next version is 1.0, so
epoch had to be used
busybox has another example (developper simply upload an package with
incorrect version):
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=97290
You can have fun time investigating which packages you have installed
has epoch in their version and why they have it, running attached perl
script ( I was boring this night ...)
Regards
#!/usr/bin/perl -w
use strict;
if ( not -e "/usr/bin/dpkg-parsechangelog" )
{
printf "You need dpkg-parsechangelog (in dpkg-dev package)";
exit 1;
}
my @packages_version=`aptitude search --disable-columns "?and(?version(:),?installed)" -F"%p"`;
my %revised;
chomp @packages_version;
my $tmpfile = "/tmp/changelog.tmp";
foreach my $package (sort @packages_version)
{
my $found=0;
if ( -e "/usr/share/doc/$package/changelog.Debian.gz" )
{
my $md5sum=`md5sum /usr/share/doc/$package/changelog.Debian.gz`;
chomp $md5sum; $md5sum=~s/\s+.*//;
if (defined ($revised{$md5sum}))
{
printf "Skipping $package, shared changelog\n"; next;
}
$revised{$md5sum}=1;
`zcat /usr/share/doc/$package/changelog.Debian.gz > $tmpfile`;
}
elsif ( -f "/usr/share/doc/$package/changelog.gz" )
{
my $md5sum=`md5sum /usr/share/doc/$package/changelog.gz`;
chomp $md5sum; $md5sum=~s/\s+.*//;
if (defined ($revised{$md5sum}))
{
printf "Skipping $package, shared changelog\n"; next;
}
$revised{$md5sum}=1;
`zcat /usr/share/doc/$package/changelog.gz > $tmpfile`
}
else
{
printf "Didn't find $package changelog\n";
next;
}
printf "Examining $package changelog \n";
open my $changelog,"$tmpfile" or die ;
my $new_version="";
my $version="";
while (<$changelog>)
{
if (/ \((\d[^)]*)\) .*urgency=/)
{
$new_version=$version;
$version=$1;
if ($new_version =~ /:/ and $version !~ /:/)
{
system (qq(dpkg-parsechangelog --from "$version" --to "$new_version" -l$tmpfile));
$found=1;
last;
}
}
}
if ($found == 0)
{
print "Didn't find version with epoch in $package changelog, maybe source package is different ...\n";
}
}
unlink "$tmpfile";
Re: Understanding versioning.
On Sun, 06 Nov 2011, Harry Putnam wrote: > Camaleón writes: > > http://www.debian.org/doc/debian-policy/ch-controlfields.html > > > epoch > > > This is a single (generally small) unsigned integer. It may be omitted, > > in which case zero is assumed. If it is omitted then the upstream_version > > may not contain any colons. > > > > It is provided to allow mistakes in the version numbers of older versions > > of a package, and also a package's previous version numbering schemes, to > > be left behind. > > What does that non-sensical sounding explanation mean? Its not as if > it is explained at the URL cited. It is a version override. For an epoch of "n", *any* version without an epoch or with an epoch that is lower than "n" will be considered *older*. So 5:0.0.1 is NEWER than 10.2.3, or than 1:9.1.4. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/2006222144.gc29...@khazad-dum.debian.net
Re: Understanding versioning.
Camaleón writes: > http://www.debian.org/doc/debian-policy/ch-controlfields.html > epoch > This is a single (generally small) unsigned integer. It may be omitted, > in which case zero is assumed. If it is omitted then the upstream_version > may not contain any colons. > > It is provided to allow mistakes in the version numbers of older versions > of a package, and also a package's previous version numbering schemes, to > be left behind. What does that non-sensical sounding explanation mean? Its not as if it is explained at the URL cited. I'm sure its perfectly reasonable if you know enough about debian generally, but if not it sounds kind of off the wall. Can you show an example of how it works? -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87pqh4rjht@newsguy.com
Re: Understanding versioning.
On Mon, 07 Nov 2011 00:38:32 +0700, Sthu Deus wrote: > When I see this aptitude note: > > [UPGRADE] libavcodec52 5:0.6.1+svn20101128-0.2squeeze2 -> 5:0.7.7-0.0 > > how do I interpret number "5:" in "5:0.6.1"? *** http://www.debian.org/doc/debian-policy/ch-controlfields.html 5.6.12 Version The version number of a package. The format is: [epoch:]upstream_version[- debian_revision] The three components here are: epoch This is a single (generally small) unsigned integer. It may be omitted, in which case zero is assumed. If it is omitted then the upstream_version may not contain any colons. It is provided to allow mistakes in the version numbers of older versions of a package, and also a package's previous version numbering schemes, to be left behind. *** HTH :-) Greetings, -- Camaleón -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/pan.2011.11.06.17.46...@gmail.com
