Issue 99899 - OOO310_m4: bogus writer2latex version
Summary: OOO310_m4: bogus writer2latex version
Status: CLOSED FIXED
Alias: None
Product: Installation
Classification: Application
Component: code (show other issues)
Version: current
Hardware: Unknown All
: P1 (highest) Trivial (vote)
Target Milestone: OOo 3.1
Assignee: Olaf Felka
QA Contact: issues@installation
URL:
Keywords:
Depends on:
Blocks: 95768
  Show dependency tree
 
Reported: 2009-03-05 00:41 UTC by rene
Modified: 2009-03-26 16:07 UTC (History)
2 users (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---


Attachments

Note You need to log in before you can comment on or make changes to this issue.
Description rene 2009-03-05 00:41:27 UTC
Issue 99417 "fixed" witer2latex packaging and in the process completels broke
its version:

-packageversion = "0.5.0.2"
+packageversion = "0.5.2"

umm, no, that's broken. The version of writer2latex *IS* 0.5.0.2. See
writer2latex' homepage (http://writer2latex.sourceforge.net/) and this:

rene@mini:~/OpenOffice.org/cws/configure24/writer2latex/download$ ls
writer2latex0502.zip

If you changed it for some upgrade or whatever reason, please do something else,
but do not break version numbers. writer2latex is external and has
a external version number which should not be changed.
Comment 1 Olaf Felka 2009-03-05 07:18:38 UTC
@ rene: Where can I find this broken version information? Please announce this
issue as a stopper at the releases list.
Comment 2 rene 2009-03-05 08:20:26 UTC
> @ rene: Where can I find this broken version information? [...]

setup_native/source/packinfo/packinfo_office.txt

> Please announce this issue as a stopper at the releases list.

Already added it as dependency of the meta-issue. This is so obviously a stopper
that I didn't bother to mail releases@
Comment 3 Olaf Felka 2009-03-05 09:02:33 UTC
@ is: Please have a look.
Comment 4 ingo.schmidt-rosbiegal 2009-03-05 09:38:20 UTC
You used the variable 

"packageversion"

that is used as version for RPMs, Solaris packages, ...

This can only contain the format:

"major.minor.micro".

So the OOo 310 m4 contains a package with the correct name:

openoffice.org3-writer2latex-0.5.2-9388.i586.rpm

"0.5.2" is the version of the package, not of the oxt file (that of course is
not influenced by the usage of "packageversion").
By the way: All other packages have the packageversion "3.1.0". 
If you do not like "0.5.2" I would suggest to use the "3.1.0". 

Additionally I changed the name to "openoffice.org3-writer2latex" instead of
"ooobasis3.1-writer2latex" because the package content is written into the brand
layer instead of the basis layer.

So I would fix this by introducing the automatically generated packageversion
"3.1.0". How do you think about this?
Comment 5 rene 2009-03-05 09:55:00 UTC
> You used the variable 
> "packageversion"
> that is used as version for RPMs, Solaris packages, ...
> This can only contain the format:
> "major.minor.micro".

Eh, what? Solaris cannot handle more digits? the rpms in m3 *have* 0.5.0.2 correct.

Anyway, then the correct fix would be 0.5.0, not 0.5.2

> So the OOo 310 m4 contains a package with the correct name:
> openoffice.org3-writer2latex-0.5.2-9388.i586.rpm

It did in m3, too (with 0.5.0.2). And 0.5.2 is NOT correct.

> "0.5.2" is the version of the package, not of the oxt file (that of course is
> not influenced by the usage of "packageversion").

Eh, yes, and? You want o claim you ship a writer2latex 0.5.2 which doesn't exist?

> By the way: All other packages have the packageversion "3.1.0". 
> If you do not like "0.5.2" I would suggest to use the "3.1.0". 

I suggest 0.5.0. Because that's what the version of writer2latex is.
writer2latex is not 3.1.0 because it's not coming from OOo, it's just
taken from Henriks site.

> Additionally I changed the name to "openoffice.org3-writer2latex" instead of
> "ooobasis3.1-writer2latex" because the package content is written into the
> brand layer instead of the basis layer.

That is correct.

> So I would fix this by introducing the automatically generated packageversion
> "3.1.0". How do you think about this?

See above. *If* you needs to scrutinize the version, just stip the .2 and use
0.5.0...
Comment 6 ingo.schmidt-rosbiegal 2009-03-05 10:09:07 UTC
> Eh, what? Solaris cannot handle more digits? the rpms in m3 *have* 0.5.0.2
correct.
> Anyway, then the correct fix would be 0.5.0, not 0.5.2

Of course the m3 contains the string 0.5.0.2, but this is not allowed. I suppose
the fourth number will simply be ignored by rpm, so that the update will not
work correctly, if you change from "0.5.0.2" to "0.5.0.x".

> Eh, yes, and? You want o claim you ship a writer2latex 0.5.2 which doesn't exist?

Well, I have to admit, that it was not a good idea to change from 0.5.0.2 to
0.5.2, which has no context to our other packages (like 0.5.0, too).

> I suggest 0.5.0. Because that's what the version of writer2latex is.
> writer2latex is not 3.1.0 because it's not coming from OOo, it's just taken
from Henriks site.
> See above. *If* you needs to scrutinize the version, just stip the .2 and use
0.5.0...

We should not mix up two different things: The version of the package, and the
version of the files it contains. All packages containing oxt files have the
version 3.1.0, although probably no oxt file has this version. The same with all
our libraries. In my opinion the oxt file version should not influence the
package version, so that "3.1.0" is the only correct package version.
Comment 7 rene 2009-03-05 12:13:09 UTC
> We should not mix up two different things: The version of the package, and the
> version of the files it contains. All packages containing oxt files have the
> version 3.1.0, although probably no oxt file has this version. The same with all

IMHO that's a bug, too. But unfortunately many of the dicts (and that's what
the oxt is you mean) have no clear version.

If the SRB etc. would be packaged, i'd argue for them to carry the correct
version, too.

>our libraries. In my opinion the oxt file version should not influence the
> package version, so that "3.1.0" is the only correct package version.

The only content of this package *is* the oxt so it does not make sense to
reflect its version.

You would have liked it if I omitted the actual extension version in
http://packages.debian.org/sid/openoffice.org-report-builder
http://packages.debian.org/sid/openoffice.org-pdfimport
etc.?
Comment 8 rene 2009-03-05 12:23:58 UTC
> The only content of this package *is* the oxt so it does not make sense to
> reflect its version.
^ not
Comment 9 rene 2009-03-05 12:28:44 UTC
(or, for this case:

http://packages.debian.org/sid/openoffice.org-writer2latex

which does not even have the OOo suffix as it's directly built out of the
external writer2latex)
Comment 10 ingo.schmidt-rosbiegal 2009-03-05 12:59:46 UTC
I really do not see any problem to have a file in version a.b.c.d in a package
with version x.y.z. There is absolutely no correlation, even if it is the only
file in the package. And there cannot even be a correlation, if packages and
files have different version schemas, like oxt files and RPMs.

And there is no need to omit the extension version. You only have to know, that
extension in version 0.5.0.2 is part of OOo 3.1.0, in which all RPM packages
have the version 3.1.0-<buildid>. This is absolutely simple and can be generated
automatically.

And believe me: Nobody realizes that he has to update the "packageversion" for
this one package in the svn module setup_native, if the version of the oxt file
was changed. This will not fit together in a very short time. 

I created cws native234 to fix this issue.

Comment 11 ingo.schmidt-rosbiegal 2009-03-05 16:09:33 UTC
Fixed by using packageversion "3.1.0" like for all other packages, too. If there
are very good reasons for another package vesion (0.5.0 ?) they should be
escalated soon.
Comment 12 Olaf Felka 2009-03-06 09:42:29 UTC
of: The related package is named
"openoffice.org3-writer2latex-3.1.0-9388.i586.rpm" now. This is consistent to
all openoffice.org3 packages.
Comment 13 ingo.schmidt-rosbiegal 2009-03-06 14:11:51 UTC
reassigning
Comment 14 Olaf Felka 2009-03-26 16:07:50 UTC
consisten naming in ooo310_m7