On Sat, 21 Aug 2010 13:51:51 -0500 Thomas Dziedzic <gostrc@gmail.com> wrote:
On Sat, Aug 21, 2010 at 1:39 PM, Adam Hani Schakaki <arch@krzd.net> wrote:
On Sat, 21 Aug 2010 13:28:32 -0500 Thomas Dziedzic <gostrc@gmail.com> wrote:
On Sat, Aug 21, 2010 at 1:25 PM, Adam Hani Schakaki <arch@krzd.net> wrote:
On Sat, 21 Aug 2010 13:18:01 -0500 Thomas Dziedzic <gostrc@gmail.com> wrote:
On Sat, Aug 21, 2010 at 12:50 PM, Adam Hani Schakaki <arch@krzd.net> wrote:
On Sun, 22 Aug 2010 01:37:33 +0800 Ray Rashif <schivmeister@gmail.com> wrote:
> On 22 August 2010 00:56, Adam Hani Schakaki <arch@krzd.net> wrote: > > I need to find a way to get the md5 hash sums. > > You just need to use makepkg -s to generate them: But of course, you > need to verify that your own download is sane, else, you will be > distributing the wrong checksums. Not to worry, as long as your > download completed fine I would say there is no need to check with an > upstream hash file. > > > -- > GPG/PGP ID: B42DDCAD Sure, this would be the normal way. But I need two different sets of hash sums. One for the 32bits files and one for the 64bits files. That's the problem.
Excerpt of the current PKGBUILD: _source_x86_64=( ${_srcurl}_64/${_ooobasis}-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-base-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-binfilter-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-calc-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-draw-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-help-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-impress-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-math-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-res-${_intver}.x86_64.rpm ${_srcurl}_64/${_ooobasis}-writer-${_intver}.x86_64.rpm ${_srcurl}_64/3.2.1/openoffice.org3-af-${_intver}.x86_64.rpm ) _source_x86=( ${_srcurl}/${_ooobasis}-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-base-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-binfilter-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-calc-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-draw-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-help-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-impress-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-math-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-res-${_intver}.i586.rpm ${_srcurl}/${_ooobasis}-writer-${_intver}.i586.rpm ${_srcurl}/3.2.1/openoffice.org3-af-${_intver}.i586.rpm )
_md5sums_x86_64=() _md5sums_x86=() source=(${_source_x86[@]}) md5sums=(${_md5sums_x86[@]})
Adam Hani Schakaki namcap -i go-openoffice-de-3.2.1-1-*.pkg.tar.xz go-openoffice-de W: Dependency included and not needed ('go-openoffice') go-openoffice-de I: Depends as namcap sees them: depends=() go-openoffice-de W: Dependency included and not needed ('go-openoffice') go-openoffice-de I: Depends as namcap sees them: depends=()
What are the differences between the two arch packages? Do they contain any arch dependent files? It seems like that since the files got different md5 hashes. They also contain binary files.
Adam Hani Schakaki
Not necessarily.
An easy way to check that is namcap. I don't know how namcap can help to check that. This is the output I get:
namcap -i go-openoffice-de-3.2.1-1-*.pkg.tar.xz go-openoffice-de W: Dependency included and not needed ('go-openoffice') go-openoffice-de I: Depends as namcap sees them: depends=() go-openoffice-de W: Dependency included and not needed ('go-openoffice') go-openoffice-de I: Depends as namcap sees them: depends=()
namcap PKGBUILD PKGBUILD (go-openoffice-de) E: Missing checksums
Adam Hani Schakaki
Hmm, I thought it was supposed to give a warning if it was an arch dependent package with no dependent on arch files. Take a look at what openoffice lang packs do (which are 'any') since I'm sure go-openoffice can do the same thing. In fact, those are the same language pack, they only go to different directories. In openoffice-i18n there is only x86 used for both architectures. So I guess I can remove the 64bits packages. But why do they create two packages if they both are the same and why have the rpms got different hashes? This is confuses me.
Adam Hani Schakaki