Re: An alternate view of PEAR QA
| From: | Lukas Smith | Date: | Tue, 20 Jan 2004 08:19:57 +0000 |
| Subject: | Re: An alternate view of PEAR QA | ||
| References: | 1 2 3 4 5 | Groups: | php.pear.qa |
| Request: | Send a blank email to pear-qa+get-387@lists.php.net to get a copy of this message | ||
Jon Parise wrote:
On Mon, Jan 19, 2004 at 09:25:28PM +0100, Lukas Smith wrote:Well no CVS means you dont have access to the history which makes the job harder and relegates QA to a pure "nagging" job.
First, the history of changes is available in the form of the release's <changelog> notes.For that changelogs need to be well maintained and even then they dont do what CVS does for you in terms of understanding how the code evolved. If you dont have CVS access you are essentially forced to rely on whatever the developer documented, which works in an ideal world, but I wonder if you need QA in an ideal world.
Second, quality assurance is not a glamorous job. It is, however, both important and rewarding.Yes. However I dont think that running tests and reporting the results is something that QA should be doing alone, nor do I think will actually motivate people do work on QA. I would really prefer a more active role when it comes time to fix issues.
I'm starting to get the feeling that my experience with software- related quality assurance groups differs from many of you here. Perhaps that explains the differing opinions.Well we are actually scary large (not quite huge) before we begun to manage alot of parts of this project. This means that we may have to take different solutions. Maybe only in the interim, maybe even in the long run. Anyways I have portrayed that opinion in various other threads. Then again I am not sure if I am the one that sees the "true" reality or that just paints bad images.
It also makes it harder to handle orphaned packages, but thats another story.
Yes, orphaned packages have little bearing on the quality of released packages. Packages become orphaned after the fact.Well it means that someone has to take over that package in order to maintain it, in case bugs are reported or feature are requested.
I'm not trying to be unrealistic, but I don't think there's another way to perform proper quality assurance.Well I fear you are. However I wish you are not :-) I dont know if trying either system will hurt in the long run. Like trying my proposal will make moving to the other system harder (eventhough its the ideal) or if your system will turn into a fiasco that will hurt PEAR and the general interest in PEAR QA. Maybe its also not that big a drama either way.
I don't think moving forward with either of those attitudes will be very productive.Yeah nevermind that.
If we're really just talking about forming a group whose job is to be on alert for potential problems that occur in released packages, it's no longer a quality assurance group; instead, it's a live maintainance group.true .. QA sounds better though :-)
Maybe we've been talking about two different things all along: I'm advocating a package-centric QA system that focuses on verifying the quality of individual releases. You're advocating a system that appears to focus on the overall quality of PEAR itself (which is why you've included plans for managing the package index and orphaned releases).Yes my system is modelled to fit itself into the current workflow and to be able to provide benefit regardless of how many people work on QA. In my past experience people seemed very reluctant to change anything in the workflow or requirements (again: test suites are not part of the PEAR package requirements, not because I didnt propose to make it such, but because majorities in various places have said no to this requirement).
We're approaching this from two opposing directions. At least that gives us a chance at meeting in the middle. =)Actually maybe we should go with your solution because its really the ideal. If it doesnt work we can still change the system. And if it works all the better!
Still, if we're not going to aggressively QA each package release, why do we need a QA team at all? It sounds all we're going to be doing is creating and empowering an entity that has the ability to fix broken package releases in the event of the package's maintainers not being available to fix problems themselves.Yes, with how PEAR works we give people a monopoly on a certain set of functionality (or atleast a given way of implementing that functionality). Currently nobody is allowed to commit anything into a package unless they have the permission of the maintainers. My proposals was aimed at defining the limits of this monopoly. So yes its not about QA in the usual sense, but its goal is definatly to improve the quality of PEAR packages, so I think the name fits. Anyways to clarify. If we were to follow one of my other suggestions and create a clean place were we only allow code that follows our regulastions, then I would definately choose a package centric preemptive QA system. But if enough people feel that we are capable to clean up the current PEAR and are willing to make this happen I am all for using such a system on the current PEAR as well. The current amounf of interest in QA has shown otherwise, but maybe they were just missing a clear concept (such as yours). regards, Lukas Smith smith@backendmedia.com _______________________________ BackendMedia www.backendmedia.com berlin@backendmedia.com Linn Zwoch Smith GbR Pariser Str. 44 D-10707 Berlin Tel +49 30 83 22 50 00 Fax +49 30 83 22 50 07