[aur-general] iceweasel md5sum problem
Hello, I'm trying to submit the web browser Iceweasel to aur, but I'm having a problem with the md5sum. I wrote the PKGBUILD, used makepkg -g to generate the md5sum and put the output into the PKGBUILD. I made the soruce package with makepkg --source and there was no problem, and doing just makepkg to build the whole package, it is possible to install iceweasel from that package without problems, etc. the only problem is that, when downloading and installing from aur, it gives an error on the integrity check. Could anyone explain me why does it happens? my PKGBUILD it's this one: http://pastebin.com/xZawmpUP Thanks in advance ;)
Jorge Barroso wrote:
Hello, I'm trying to submit the web browser Iceweasel to aur, but I'm having a problem with the md5sum. I wrote the PKGBUILD, used makepkg -g to generate the md5sum and put the output into the PKGBUILD. I made the soruce package with makepkg --source and there was no problem, and doing just makepkg to build the whole package, it is possible to install iceweasel from that package without problems, etc. the only problem is that, when downloading and installing from aur, it gives an error on the integrity check. Could anyone explain me why does it happens? my PKGBUILD it's this one:
Thanks in advance ;)
You are using arrays for many fields which shouldn't be arrays like: pkgname, pkgver, pkgrel. You should also quote the source URL.
to, 2012-05-10 kello 19:09 +0200, Jorge Barroso kirjoitti: pkgname pkgver pkgrel pkgdesc url none of those are arrays, so remove (). They might confuse makepkg, those ' marks are undeed tooas none of them include spaces. pkgdesc being an exeption, " or ' is needed there. Not sure, but having md5sums before source could be a culprit, anyway I would put them after the sources. optdepends usually has format optdepends=('pkgname: small description of thefunctionallity it adds') tar xf $pkgname-$pkgver-$pkgrel-$arch.pkg.tar.xz is not needed, pacman can extract pretty much all kind of achives on its own .tar.{gz,bz2,xz} deb rpm zip, files atleast. The tar xf command was needed in the manga app PKGBUILD becouse it shipped only in deb format and deb archives have 2 archives in it, one with the data and one with meta-data files and pacman only extracts files in source=() [The main deb file] One usually should not upload a binary packages into AUR, but instead it should be built from source code. Of course in case of s HUGE application that takes a lot of time to compile it is nice to have a binary version too. Or in case like I had a ?year? ago. Back then we didn't ship sdk package for the go-openoffice and it ment that one would have to rebuuild it from source just to be able to compile a 100kb addon for it to support Finnish spellcheckking (voikko) so I also made -bin version from Ubuntu binary packages. I'm quite sure one should go totally with the parabola Linux instead of Arch if libre is thing they want. We have a lots of stuff that go against libre in the core repos, and just gettinga libre browser is totally ridicilous. You also are missing the i686 version. If you really think a libre webrowser on non libre system is a good idea you should be simply stealing the PKGBUILD from here https://projects.parabolagnulinux.org/abslibre.git/tree/libre/iceweasel-libr...
2012/5/10 Jorge Barroso <jorge.barroso.11@gmail.com>:
Hello, I'm trying to submit the web browser Iceweasel to aur, but I'm having a problem with the md5sum. I wrote the PKGBUILD, used makepkg -g to generate the md5sum and put the output into the PKGBUILD. I made the soruce package with makepkg --source and there was no problem, and doing just makepkg to build the whole package, it is possible to install iceweasel from that package without problems, etc. the only problem is that, when downloading and installing from aur, it gives an error on the integrity check. Could anyone explain me why does it happens? my PKGBUILD it's this one:
Thanks in advance ;)
The md5sums in the pastebin is different from the AUR. Correct one is in the pastebin. That's the answer for your issue, but +1 to the comments made by others. Rafael
2012/5/10 A Rojas <nqn1976list@gmail.com>
You are using arrays for many fields which shouldn't be arrays like: pkgname, pkgver, pkgrel. You should also quote the source URL.
------------------------------------------------------------------------------------------------------------------ 2012/5/10 Jesse Juhani Jaara <jesse.jaara@gmail.com>
to, 2012-05-10 kello 19:09 +0200, Jorge Barroso kirjoitti:
pkgname pkgver pkgrel pkgdesc url
none of those are arrays, so remove (). They might confuse makepkg, those ' marks are undeed tooas none of them include spaces. pkgdesc being an exeption, " or ' is needed there.
Not sure, but having md5sums before source could be a culprit, anyway I would put them after the sources.
optdepends usually has format optdepends=('pkgname: small description of thefunctionallity it adds')
tar xf $pkgname-$pkgver-$pkgrel-$arch.pkg.tar.xz is not needed, pacman can extract pretty much all kind of achives on its own .tar.{gz,bz2,xz} deb rpm zip, files atleast. The tar xf command was needed in the manga app PKGBUILD becouse it shipped only in deb format and deb archives have 2 archives in it, one with the data and one with meta-data files and pacman only extracts files in source=() [The main deb file]
One usually should not upload a binary packages into AUR, but instead it should be built from source code. Of course in case of s HUGE application that takes a lot of time to compile it is nice to have a binary version too. Or in case like I had a ?year? ago. Back then we didn't ship sdk package for the go-openoffice and it ment that one would have to rebuuild it from source just to be able to compile a 100kb addon for it to support Finnish spellcheckking (voikko) so I also made -bin version from Ubuntu binary packages.
I'm quite sure one should go totally with the parabola Linux instead of Arch if libre is thing they want. We have a lots of stuff that go against libre in the core repos, and just gettinga libre browser is totally ridicilous. You also are missing the i686 version.
If you really think a libre webrowser on non libre system is a good idea you should be simply stealing the PKGBUILD from here
https://projects.parabolagnulinux.org/abslibre.git/tree/libre/iceweasel-libr...
-------------------------------------------------------------------------------------------------------------------------
2012/5/10 rafael ff1 <rafael.f.f1@gmail.com>
2012/5/10 Jorge Barroso <jorge.barroso.11@gmail.com>:
Hello, I'm trying to submit the web browser Iceweasel to aur, but I'm having a problem with the md5sum. I wrote the PKGBUILD, used makepkg -g to generate the md5sum and put the output into the PKGBUILD. I made the soruce package with makepkg --source and there was no problem, and doing just makepkg to build the whole package, it is possible to install iceweasel from that package without problems, etc. the only problem is that, when downloading and installing from aur, it gives an error on the integrity check. Could anyone explain me why does it happens? my PKGBUILD it's this one:
Thanks in advance ;)
The md5sums in the pastebin is different from the AUR. Correct one is in the pastebin. That's the answer for your issue, but +1 to the comments made by others.
Rafael
-------------------------------------------------------------------------------------------------------------------------------- Thanks to all of you, those are my first packages as you should know and I'm using the proto of /usr/share/pacman to make the pkgbuild, so I'm using those arrays I think I can use for any package, but I didn't know that () will confuse makepkg (so sensitive...), I'll try to correct it. The md5sums in the pastebin is different because it has been changing between two different outputs anytime I did makepkg -g, but even now that it is not changing, and the makepkg -g output is ever the same, it stills not accepting it. Jesse, I'll try then to use that PKGBUILD, I looked for it, because I suposed it should exist, but as I didn't find it, I downloaded the package and modified the .PKGINFO file. Thanks for your help with that PKGBUILD. I know archlinux is a non libre system, and that's not bad at all because, for example, flash it's very usefull nowadays, because gnash has a lot of things to improve to be a serious subtitute, but, ever I can, I try to use free software (if it has not great problems, like gnash taking my earlier example), and Iceweasel with iceweasel-sync it's as good and fast as firefox, so... I liked it and thought about submitting it :)
2012/5/10 Jorge Barroso <jorge.barroso.11@gmail.com>:
2012/5/10 Jorge Barroso <jorge.barroso.11@gmail.com>:
Hello, I'm trying to submit the web browser Iceweasel to aur, but I'm having a problem with the md5sum. I wrote the PKGBUILD, used makepkg -g to generate the md5sum and put the output into the PKGBUILD. I made the soruce package with makepkg --source and there was no problem, and doing just makepkg to build the whole package, it is possible to install iceweasel from that package without problems, etc. the only problem is that, when downloading and installing from aur, it gives an error on the integrity check. Could anyone explain me why does it happens? my PKGBUILD it's this one:
Thanks in advance ;)
The md5sums in the pastebin is different from the AUR. Correct one is in the pastebin. That's the answer for your issue, but +1 to the comments made by others.
Rafael
--------------------------------------------------------------------------------------------------------------
The md5sums in the pastebin is different because it has been changing between two different outputs anytime I did makepkg -g, but even now that it is not changing, and the makepkg -g output is ever the same, it stills not accepting it.
What? The source is the same, stored on a server, therefore its md5sum WILL NOT change everytime you run 'makepkg -g' unless you are running 'makepkg -g' for different sources or your source is corrupted. In the last case, remove and download again the source. $ md5sum iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz 638379a2dcf1f3d116cf92a06c8ee9b9 iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz Rafael
2012/5/10 rafael ff1 <rafael.f.f1@gmail.com>
What? The source is the same, stored on a server, therefore its md5sum WILL NOT change everytime you run 'makepkg -g' unless you are running 'makepkg -g' for different sources or your source is corrupted. In the last case, remove and download again the source.
$ md5sum iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz 638379a2dcf1f3d116cf92a06c8ee9b9 iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz
Rafael
Jaja yeah I though the same, but sometimes it gaves that md5sum and another times it gaves a different checksum, finished on 6o6 I think (i don't remember the exactly output)
On 2012-05-10 15:13, Jorge Barroso wrote:
2012/5/10 rafael ff1 <rafael.f.f1@gmail.com>
What? The source is the same, stored on a server, therefore its md5sum WILL NOT change everytime you run 'makepkg -g' unless you are running 'makepkg -g' for different sources or your source is corrupted. In the last case, remove and download again the source.
$ md5sum iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz 638379a2dcf1f3d116cf92a06c8ee9b9 iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz
Rafael
Jaja yeah I though the same, but sometimes it gaves that md5sum and another times it gaves a different checksum, finished on 6o6 I think (i don't remember the exactly output)
Just in case, a diferent arch will result in a different package and therefore a different checksum. Is there any reason you've made this package amd64 only? -- Hugo Osvaldo Barrera
2012/5/10 Hugo Osvaldo Barrera <hugo@osvaldobarrera.com.ar>:
On 2012-05-10 15:13, Jorge Barroso wrote:
2012/5/10 rafael ff1 <rafael.f.f1@gmail.com>
What? The source is the same, stored on a server, therefore its md5sum WILL NOT change everytime you run 'makepkg -g' unless you are running 'makepkg -g' for different sources or your source is corrupted. In the last case, remove and download again the source.
$ md5sum iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz 638379a2dcf1f3d116cf92a06c8ee9b9 iceweasel-libre-12.0.3-1-x86_64.pkg.tar.xz
Rafael
Jaja yeah I though the same, but sometimes it gaves that md5sum and another times it gaves a different checksum, finished on 6o6 I think (i don't remember the exactly output)
Just in case, a diferent arch will result in a different package and therefore a different checksum.
His package in AUR: 3aca2484df769ef5a44212395c1c5937 64-bit package: 638379a2dcf1f3d116cf92a06c8ee9b9 32-bit package: 47443a0a73f9a53ba6d0c007a213b8b5
Is there any reason you've made this package amd64 only?
-- Hugo Osvaldo Barrera
On Thu, May 10, 2012 at 09:00:45PM -0300, rafael ff1 wrote: <snip>
His package in AUR: 3aca2484df769ef5a44212395c1c5937 64-bit package: 638379a2dcf1f3d116cf92a06c8ee9b9 32-bit package: 47443a0a73f9a53ba6d0c007a213b8b5
just for reference, which usually avoids yet more questions: [1] if [[ "$CARCH" == "x86_64" ]]; then md5sum=(...) else md5sum=(...) fi working examples are, as usual, out there on the AUR. cheers! mar77i [1] https://wiki.archlinux.org/index.php/PKGBUILD#arch
2012/5/10 Martti Kühne <mysatyre@gmail.com>:
On Thu, May 10, 2012 at 09:00:45PM -0300, rafael ff1 wrote: <snip>
His package in AUR: 3aca2484df769ef5a44212395c1c5937 64-bit package: 638379a2dcf1f3d116cf92a06c8ee9b9 32-bit package: 47443a0a73f9a53ba6d0c007a213b8b5
just for reference, which usually avoids yet more questions: [1] if [[ "$CARCH" == "x86_64" ]]; then md5sum=(...) else md5sum=(...) fi
working examples are, as usual, out there on the AUR.
cheers! mar77i
+1 to this solution, as AUR web interface parse it perfectly according to my architecture, showing the correct depends - which does not happen with source=(foobar) [ "$CARCH" == "i686" ] && source=(lib32-foobar) Rafael
source=(foobar) [ "$CARCH" == "i686" ] && source=(lib32-foobar)
ops, it seems I #fail. The correct would be: source=(foobar) [ "$CARCH" == "x86_64" ] && source=(lib32-foobar)
From https://aur.archlinux.org/packages.php?ID=50893 linux-mainline 3.4rc6-1
Do you guys have any problem with hibernating? my computer cannot boot after running pm-hibernate :L The only solution for this is to unplug the power source wait a few seconds and the plug again the power source cable. After this, when you boot the computer there is a warning message from the bios about incorrect setting of the cpu frequency. Then I am prompted to the bios configuration screen but all settings look ok. After restart the computer boots normally and resume successfully from hibernating. Please mention any lofg file I can attach so we can find the cause of this problem.
On Thu, May 10, 2012 at 07:55:00PM +0200, Jorge Barroso wrote:
Thanks to all of you, those are my first packages as you should know and I'm using the proto of /usr/share/pacman to make the pkgbuild, so I'm using those arrays I think I can use for any package, but I didn't know that () will confuse makepkg (so sensitive...), I'll try to correct it.
actually, bash syntax is something to be really careful with if one isn't too familiar with it. [1] gives you some funny examples about it, I also recommend to read [2] - it's a great read and will help you greatly to get work done in the shell cli without having to rely on GUIs all the time. Also, this: martti@deepthought:~$ e=(abc) martti@deepthought:~$ echo $e abc In our case there is another possible source of errors, which is makepkg, and if it would actually break on something it may be considered a bug in some, but it might simply turn out to be a non-trivial edge case in other cases. I'd actually find it interesting to figure out more of this just to see if I can break it... :)
The md5sums in the pastebin is different because it has been changing between two different outputs anytime I did makepkg -g, but even now that it is not changing, and the makepkg -g output is ever the same, it stills not accepting it.
Something odd may be going on with your hardware or os. Not sure where, but if memtest doesn't find errors, consider checking your disk with smartmontools and try rkhunter, just in case. md5 is supposed to be an empirical checksum, not a dice roll.
Jesse, I'll try then to use that PKGBUILD, I looked for it, because I suposed it should exist, but as I didn't find it, I downloaded the package and modified the .PKGINFO file. Thanks for your help with that PKGBUILD. I know archlinux is a non libre system, and that's not bad at all because, for example, flash it's very usefull nowadays, because gnash has a lot of things to improve to be a serious subtitute, but, ever I can, I try to use free software (if it has not great problems, like gnash taking my earlier example), and Iceweasel with iceweasel-sync it's as good and fast as firefox, so... I liked it and thought about submitting it :)
just fyi, the mozilla license is listed as being less restrictive than gnu gpl. [3] cheers! mar77i [1] http://mywiki.wooledge.org/BashParser [2] http://mywiki.wooledge.org/BashGuide [3] http://en.wikipedia.org/wiki/Comparison_of_free_and_open_source_software_lic...
participants (7)
-
A Rojas
-
Hugo Osvaldo Barrera
-
Jesse Juhani Jaara
-
Jorge Barroso
-
Martti Kühne
-
rafael ff1
-
Sébastien le Preste de Vauban