[aur-dev] lib32-gtk3 block and no AUR User Guidelines presidence
The maintainer of lib32-gtk3, alucryd, recently updated the version to 3.16.1 although it's dependencies are in testing. So now anyone wanting to install something depending on lib32-gtk3 would now be forced to enable the testing repository. In alucryd's comments with another user he seems to be under the impression all users already have lib32-gtk3 installed, and he isn't forcing anyone to update. He is forgetting all of the people that are installing something for the first time that depends on lib32-gtk. They are completely blocked without adding the testing repository. I reviewed the AUR User Guidelines and can find nothing to prevent this type of lapse in logic. This isn't the first time I have seen this issue. Since there isn't a separate testing AUR, can something be added to the AUR User Guidelines maybe stating packages that require the testing repository must have "-testing" added to the name or whatever would best resolve this problem. Thank You For Your Time, Curtis Lee Bolin
As far as I was aware, it was a (pretty well-respected) gentlemen's rule to append '-testing' or '-unstable' onto any package which is in testing, is unstable, or pulls from any of the official testing repos. Perhaps I am wrong. Something to note is that an AUR search for packages containing '-testing' in the name returns 17 packages and '-unstable' 44. I should think it would be general courtesy for any maintainer to hold off on an update like this but I am aware some parts of the community may be allergic to this thinking. Bear in mind that the AUR guidelines are on the wiki, so anyone can make this change if the general consensus is that the community agrees with you. -- Four word witty remark
This discussion should be on aur-general. Aur-dev is for discussion on the development of the actually Aur website not the packages it hosts. Sent from my iPhone
On Apr 9, 2015, at 7:16 PM, David Phillips <dbphillipsnz@gmail.com> wrote:
As far as I was aware, it was a (pretty well-respected) gentlemen's rule to append '-testing' or '-unstable' onto any package which is in testing, is unstable, or pulls from any of the official testing repos. Perhaps I am wrong. Something to note is that an AUR search for packages containing '-testing' in the name returns 17 packages and '-unstable' 44.
I should think it would be general courtesy for any maintainer to hold off on an update like this but I am aware some parts of the community may be allergic to this thinking.
Bear in mind that the AUR guidelines are on the wiki, so anyone can make this change if the general consensus is that the community agrees with you.
-- Four word witty remark
Am 09.04.2015 um 21:09 schrieb Curtis Lee Bolin:
The maintainer of lib32-gtk3, alucryd, recently updated the version to 3.16.1 although it's dependencies are in testing. So now anyone wanting to install something depending on lib32-gtk3 would now be forced to enable the testing repository.
He warned about this in an AUR comments [1], that is fair enough IMHO.
In alucryd's comments with another user he seems to be under the impression all users already have lib32-gtk3 installed, and he isn't forcing anyone to update. He is forgetting all of the people that are installing something for the first time that depends on lib32-gtk. They are completely blocked without adding the testing repository.
He also gave a solution for this issue.
I reviewed the AUR User Guidelines and can find nothing to prevent this type of lapse in logic. This isn't the first time I have seen this issue. Since there isn't a separate testing AUR, can something be added to the AUR User Guidelines maybe stating packages that require the testing repository must have "-testing" added to the name or whatever would best resolve this problem.
The AUR User Guidelines are about AUR, not about official packages.
Thank You For Your Time, Curtis Lee Bolin
Best Regards Stefan Husmann [1] Updated to 3.16.1 (still in [testing]). This brings 2 new deps, lib32-libepoxy and lib32-libcanberra-gtk3. The latter is a terrible terrible circular dep, it should depend on lib32-gtk3, but I only put it in makedeps since lib32-libcanberra in [multilib] seems to use the same approach.
participants (4)
-
Curtis Lee Bolin
-
Daniel Wallace
-
David Phillips
-
Stefan Husmann