[aur-general] AUR package updating enhancements
Hello, following the weird titled "AUR webpage" thread ( http://www.archlinux.org/pipermail/aur-general/2009-July/thread.html#5602 ) I think the way maintainers handle their packages could have some things simplified. I agree that simple updates to PKGBUILDs should be easier. I've seen many packages with typos in PKGBUILDs, showing that the extra workload of having to archive the files and upload them doesn't necessary make users test things before uploading. That's all up to the maintainer's will to do things right. I think having a simple syntax highlighted code editor to quick edit the PKGBUILD or at least a way to simply update the PKGBUILD without having to archive and upload everything every time would be a great help. Some alternatives google gave me, dunno if there are any better: http://alexgorbatchev.com/wiki/SyntaxHighlighter http://codepress.sourceforge.net/ http://marijn.haverbeke.nl/codemirror/ http://helene.muze.nl/ What do you think?
2009/8/26 Nélson «VuDu» Cunha <vudu.curse@gmail.com>:
Hello, following the weird titled "AUR webpage" thread ( http://www.archlinux.org/pipermail/aur-general/2009-July/thread.html#5602 ) I think the way maintainers handle their packages could have some things simplified.
I agree that simple updates to PKGBUILDs should be easier. I've seen many packages with typos in PKGBUILDs, showing that the extra workload of having to archive the files and upload them doesn't necessary make users test things before uploading. That's all up to the maintainer's will to do things right.
I think having a simple syntax highlighted code editor to quick edit the PKGBUILD or at least a way to simply update the PKGBUILD without having to archive and upload everything every time would be a great help.
Some alternatives google gave me, dunno if there are any better: http://alexgorbatchev.com/wiki/SyntaxHighlighter http://codepress.sourceforge.net/ http://marijn.haverbeke.nl/codemirror/ http://helene.muze.nl/
What do you think?
Nah, doesn't seem very kiss to me. IMO if you want something easier to use create a tool to allow uploading from the cli. Ronald
2009/8/26 Nélson «VuDu» Cunha <vudu.curse@gmail.com>:
I agree that simple updates to PKGBUILDs should be easier. I've seen many packages with typos in PKGBUILDs, showing that the extra workload of having to archive the files and upload them doesn't necessary make users test things before uploading. That's all up to the maintainer's will to do things right.
What do you think?
I think this is a poor argumentation. "some maintainers are already screwing things up, so there is no problem with encouraging them / making all this easier for them" The fact is, maintainers should always test their changes, even the most trivial ones. Besides, with the existing AUR helpers [http://www.archlinux.it/wiki/index.php?title=AUR_Helpers] , it should make all this very easy : - download / extract current package - edit pkgbuild - run makepkg and namcap - makepkg --source and upload you could even automate all these steps easily I think.
On Wed, Aug 26, 2009 at 03:45:56PM +0200, Xavier wrote:
Besides, with the existing AUR helpers [http://www.archlinux.it/wiki/index.php?title=AUR_Helpers] , it should make all this very easy : - download / extract current package - edit pkgbuild - run makepkg and namcap - makepkg --source and upload
you could even automate all these steps easily I think.
yes, try pkgman: http://aur.archlinux.org/packages.php?ID=17100 it does this all! oh, i have to add it to the wiki... --
On Wed, Aug 26, 2009 at 4:35 PM, vlad<vla@uni-bonn.de> wrote:
On Wed, Aug 26, 2009 at 03:45:56PM +0200, Xavier wrote:
Besides, with the existing AUR helpers [http://www.archlinux.it/wiki/index.php?title=AUR_Helpers] , it should make all this very easy : - download / extract current package - edit pkgbuild - run makepkg and namcap - makepkg --source and upload
you could even automate all these steps easily I think.
yes, try pkgman: http://aur.archlinux.org/packages.php?ID=17100 it does this all! oh, i have to add it to the wiki...
good to know :) it would indeed be nice to have it in the wiki btw, the link I gave is completely wrong and broken, sorry for that. it should have been : http://wiki.archlinux.org/index.php/AUR_Helpers (for some reason, looking for "archwiki aur tools" with google french only returns the bad link on the first page, and the good link is nowhere to be found. but it is probably a bad idea to use google in the first place for this kind of search)
Thanks for the link, some sure help. I just knew a couple of them. ;) On Wed, Aug 26, 2009 at 15:53, Xavier<shiningxc@gmail.com> wrote:
On Wed, Aug 26, 2009 at 4:35 PM, vlad<vla@uni-bonn.de> wrote:
On Wed, Aug 26, 2009 at 03:45:56PM +0200, Xavier wrote:
Besides, with the existing AUR helpers [http://www.archlinux.it/wiki/index.php?title=AUR_Helpers] , it should make all this very easy : - download / extract current package - edit pkgbuild - run makepkg and namcap - makepkg --source and upload
you could even automate all these steps easily I think.
yes, try pkgman: http://aur.archlinux.org/packages.php?ID=17100 it does this all! oh, i have to add it to the wiki...
good to know :) it would indeed be nice to have it in the wiki
btw, the link I gave is completely wrong and broken, sorry for that. it should have been : http://wiki.archlinux.org/index.php/AUR_Helpers
(for some reason, looking for "archwiki aur tools" with google french only returns the bad link on the first page, and the good link is nowhere to be found. but it is probably a bad idea to use google in the first place for this kind of search)
Sorry, but please don't try to make it sound like I said that because that's not an argument at all. Following what I said before, making steps harder isn't necessarily going to make people walk through them more carefully. I don't agree that an on-line code editor would make people do more mistakes, because good maintainers would continue to test the PKGBUILD before making the changes on the on-line editor, or just simply not using it. An editor would just save the pain to archive and upload (not testing them) simple changes to the PKGBUILD, that make probably the biggest part of AUR packages updates. Another thing, not just related to the PKGBUILD itself, is that the update process is totally unintuitive. For any kind of update to a package, being it major or minor, one has to always replace the package with a new one. That's not intuitive and... doesn't KISS also support that "less is more"? Personally I don't buy the "make it hard so that they (the few that don't quit) make it better", but that's just an opinion... and everyone has it's own. Must respect that. ;) On Wed, Aug 26, 2009 at 14:45, Xavier<shiningxc@gmail.com> wrote:
2009/8/26 Nélson «VuDu» Cunha <vudu.curse@gmail.com>:
I agree that simple updates to PKGBUILDs should be easier. I've seen many packages with typos in PKGBUILDs, showing that the extra workload of having to archive the files and upload them doesn't necessary make users test things before uploading. That's all up to the maintainer's will to do things right.
What do you think?
I think this is a poor argumentation. "some maintainers are already screwing things up, so there is no problem with encouraging them / making all this easier for them" The fact is, maintainers should always test their changes, even the most trivial ones.
Besides, with the existing AUR helpers [http://www.archlinux.it/wiki/index.php?title=AUR_Helpers] , it should make all this very easy : - download / extract current package - edit pkgbuild - run makepkg and namcap - makepkg --source and upload
you could even automate all these steps easily I think.
On 27/ago/09, at 01:35, Nélson «VuDu» Cunha wrote:
Sorry, but please don't try to make it sound like I said that because that's not an argument at all. Following what I said before, making steps harder isn't necessarily going to make people walk through them more carefully. I don't agree that an on-line code editor would make people do more mistakes, because good maintainers would continue to test the PKGBUILD before making the changes on the on-line editor, or just simply not using it. An editor would just save the pain to archive and upload (not testing them) simple changes to the PKGBUILD, that make probably the biggest part of AUR packages updates.
I don't understand why the developers should write, test and deploy new code in the website to perform a task that can already be done with CLI tools. Which is exactly "the pain to archive and upload"? makepkg--source && $AUR_HELPER *.src.tar.gz isn't much of pain to me :) Another issue that comes to my mind: how would you update the md5sums with the online editor?
On Thu 27 Aug 2009 01:47 +0200, Arkham wrote:
On 27/ago/09, at 01:35, Nélson «VuDu» Cunha wrote:
Sorry, but please don't try to make it sound like I said that because that's not an argument at all. Following what I said before, making steps harder isn't necessarily going to make people walk through them more carefully. I don't agree that an on-line code editor would make people do more mistakes, because good maintainers would continue to test the PKGBUILD before making the changes on the on-line editor, or just simply not using it. An editor would just save the pain to archive and upload (not testing them) simple changes to the PKGBUILD, that make probably the biggest part of AUR packages updates.
I don't understand why the developers should write, test and deploy new code in the website to perform a task that can already be done with CLI tools. Which is exactly "the pain to archive and upload"? makepkg--source && $AUR_HELPER *.src.tar.gz isn't much of pain to me :) Another issue that comes to my mind: how would you update the md5sums with the online editor?
Yeah this was already brought up, not too long ago. An editor will not be implemented for the AUR. There are already many capable editors available such as vim, or emacs. Please choose one of them. There's no point in adding an ad hoc informally-specified bug-ridden slow implementation of half of a text editor to the AUR. You're lucky enough to get a textbox for comments. ;) Good luck!
Ok, five consecutive negative replies ought to be enough, even if some of them look like they just read the last reply before them and simply missed my point. If I would've start this in the forums instead and maybe I could get a couple of neutral or even positive (in a constructive way) comments, who knows? Clearly my fault... this is a mailing list after all! Still, thanks for the replies ;) On Thu, Aug 27, 2009 at 01:20, Loui Chang<louipc.ist@gmail.com> wrote:
On Thu 27 Aug 2009 01:47 +0200, Arkham wrote:
On 27/ago/09, at 01:35, Nélson «VuDu» Cunha wrote:
Sorry, but please don't try to make it sound like I said that because that's not an argument at all. Following what I said before, making steps harder isn't necessarily going to make people walk through them more carefully. I don't agree that an on-line code editor would make people do more mistakes, because good maintainers would continue to test the PKGBUILD before making the changes on the on-line editor, or just simply not using it. An editor would just save the pain to archive and upload (not testing them) simple changes to the PKGBUILD, that make probably the biggest part of AUR packages updates.
I don't understand why the developers should write, test and deploy new code in the website to perform a task that can already be done with CLI tools. Which is exactly "the pain to archive and upload"? makepkg--source && $AUR_HELPER *.src.tar.gz isn't much of pain to me :) Another issue that comes to my mind: how would you update the md5sums with the online editor?
Yeah this was already brought up, not too long ago. An editor will not be implemented for the AUR. There are already many capable editors available such as vim, or emacs. Please choose one of them.
There's no point in adding an ad hoc informally-specified bug-ridden slow implementation of half of a text editor to the AUR.
You're lucky enough to get a textbox for comments. ;)
Good luck!
Nélson «VuDu» Cunha <vudu.curse@gmail.com> wrote:
Ok, five consecutive negative replies ought to be enough, even if some of them look like they just read the last reply before them and simply missed my point. If I would've start this in the forums instead and maybe I could get a couple of neutral or even positive (in a constructive way) comments, who knows? Clearly my fault... this is a mailing list after all!
Still, thanks for the replies ;)
After having read the previous discussion in which I was initially for the idea before some good points were made against it, I no longer have a strong opinion one way or the other. Having said that, I will add that I think it's unfortunate that so many people have a visceral negative reaction without truly considering the benefits. Some replies clearly show misunderstandings that would not have arisen if the idea had been approached with an open mind (e.g. the need to handle md5sums*, the "complexity" of "implementing" an editor**). I'm not saying that this would have led to a different outcome, but it would have been nicer to have a neutral conversation about it rather than several posts throwing everything they could think of at the idea, including points which don't even apply. It's like suggesting to go to the beach and instead of someone objecting with "I think it's too cold today... have you checked the weather?" you get a 5 minute tirade about much it sucks to get sand in your clothes, how swallowing seawater is bad for you, that UV radiation is bad for the skin, that driving to the beach costs money and creates pollution, that people die in tsunamis on beaches, that you can be eaten by sharks, etc... which all boil down to the objector really saying "I haven't left the house in ages and I don't wanna.". * The PKGBUILD itself is not md5sum'd. ** It would be a simple textarea. Wtf said anything about keyboard shortcuts and syntax highlighting? (Again, I'm not supporting the idea, just criticizing the grumpy-old-man approach to the discussion.)
On Thu, Aug 27, 2009 at 7:20 PM, Xyne<xyne@archlinux.ca> wrote:
[...] It's like suggesting to go to the beach and instead of someone objecting with "I think it's too cold today... have you checked the weather?" you get a 5 minute tirade about much it sucks to get sand in your clothes, how swallowing seawater is bad for you, that UV radiation is bad for the skin, that driving to the beach costs money and creates pollution, that people die in tsunamis on beaches, that you can be eaten by sharks, etc... which all boil down to the objector really saying "I haven't left the house in ages and I don't wanna.".
LOL to be honest, those are all valid points you mentioned. I hate going to the beach in the summer for some of the points you mentioned, though going there when it storms is great! anyway...clearly not the point you wanted to make
* The PKGBUILD itself is not md5sum'd. ** It would be a simple textarea. Wtf said anything about keyboard shortcuts and syntax highlighting?
(Again, I'm not supporting the idea, just criticizing the grumpy-old-man approach to the discussion.)
yes we're getting old but I still don't like it. I like the AUR helpers approach though. I really don't see the point changing the PKGBUILD in the terminal (or finding md5sum or whatever), go to you webbrowser, start the editor there and change it. You could even make a typo rendering the PKGBUILD not functioning (yes you could test it by downloading the changed PKGBUILD from AUR but well...you know). But in every discussion someone can always change my mind with good arguments... Ronald
On Thu 27 Aug 2009 19:20 +0200, Xyne wrote:
It's like suggesting to go to the beach and instead of someone objecting with "I think it's too cold today... have you checked the weather?" you get a 5 minute tirade about much it sucks to get sand in your clothes, how swallowing seawater is bad for you, that UV radiation is bad for the skin, that driving to the beach costs money and creates pollution, that people die in tsunamis on beaches, that you can be eaten by sharks, etc... which all boil down to the objector really saying "I haven't left the house in ages and I don't wanna.".
It's like mailing a mini clone of myself to China with a letter I want to write in mind and having him write the letter there, rather than just writing the letter myself and putting that in the mail instead. Plus he's not as smart as me, so he will take longer to write it and might make some mistakes that I wouldn't.
On Thu, Aug 27, 2009 at 7:20 PM, Xyne<xyne@archlinux.ca> wrote:
It's like suggesting to go to the beach and instead of someone objecting with "I think it's too cold today... have you checked the weather?" you get a 5 minute tirade about much it sucks to get sand in your clothes, how swallowing seawater is bad for you, that UV radiation is bad for the skin, that driving to the beach costs money and creates pollution, that people die in tsunamis on beaches, that you can be eaten by sharks, etc... which all boil down to the objector really saying "I haven't left the house in ages and I don't wanna.".
So my suggestion would be to go on a rock beach, with a snorkel, an umbrella, a lot of sunscreen, going by bike or by foot, checking the forecast to avoid tsunami, and avoid beaches populated by sharks. It requires some more work and put additional constraints, but it still achieves your goal, in a safer/better way :) (now you can try to make the link with my real reply earlier in that thread..)
Nélson «VuDu» Cunha wrote:
Sorry, but please don't try to make it sound like I said that because that's not an argument at all. Following what I said before, making steps harder isn't necessarily going to make people walk through them more carefully. I don't agree that an on-line code editor would make people do more mistakes, because good maintainers would continue to test the PKGBUILD before making the changes on the on-line editor, or just simply not using it. An editor would just save the pain to archive and upload (not testing them) simple changes to the PKGBUILD, that make probably the biggest part of AUR packages updates.
Another thing, not just related to the PKGBUILD itself, is that the update process is totally unintuitive. For any kind of update to a package, being it major or minor, one has to always replace the package with a new one. That's not intuitive and... doesn't KISS also support that "less is more"?
Personally I don't buy the "make it hard so that they (the few that don't quit) make it better", but that's just an opinion... and everyone has it's own. Must respect that. ;)
Now, if it was said "patches welcome", would any ever be forthcoming? Thats like rejecting the idea without rejecting it... :) Allan
participants (9)
-
Allan McRae
-
Arkham
-
Loui Chang
-
Nélson «VuDu» Cunha
-
Ronald van Haren
-
Ronald van Haren
-
vlad
-
Xavier
-
Xyne