search

nvdaaddons / reviews

An issue-only repo for bringing add-ons outside of official community add-ons website to wider audiences in hopes of bringing these add-ons to the community add-ons website
0 stars 0 forks source link

Design form for submitting reviews #2

Open derekriemer opened 6 years ago

derekriemer commented 6 years ago

https://gist.github.com/bmcbride/62600e48274961819084

We could potentially pole this off. I have a couple of problems to investigate though.

  1. Our users will submit here, but they may not know this goes to github.
  2. The user won't receive notifications like they do for creating a github repo in the first place.

A couple of thoughts on how to solve this.

  1. require one of two options: Email address or github username.
  2. Just say that a github username is a requirement for addon reviews.

We need to explain that not having a github username will require a team member to manually interact with the user over email, which will both decrease transparency of the review, and make it more likely that a reviewer will forget to send out the email. I am tempted to require a github username for this reason. I don't want to have to blast out emails and remember or seperately note the status of reviews, since github can help us with this. Also, reviewers won't be as able to comment on their review if they don't have a github username. Now we have the public problem which we had on the mailing list, although we have solved the problem of addon authors not wanting to subscribe to nvda-addons and deal with users. My goal is to make this as simple as possible for reviewers and reviewees.

Thoughts @nvaccess @jcsteh @josephsl and other stakeholders?

nvdaes commented 6 years ago

A couple of thoughts on how to solve this.

  1. require one of two options: Email address or github username.
  2. Just say that a github username is a requirement for addon reviews.

I think a GitHub username shouldn't be required, since the purpose of the form could be to avoid precissely that users need a GitHub account to open an issue. Anyway, sometimes I think users should use a GitHub repo for add-ons to allow diffs review. In the other hand, I like the maximum flexibility, so really I don't know if this has a solution. Really the first problem, imo, is that the number of reviewers is very small, and also, who can say that a reviewer is or not qualified? Just thinking in a short of brainstone :)

jcsteh commented 6 years ago

I tend to think we should just require people to have a GitHub account:

  1. It's really not that hard to create one.
  2. The alternative of using email rather defeats the point of having a GitHub issue to track the review in the first place.
  3. For better or for worse (mostly for better I think), we're starting to centralise on GitHub for all NVDA stuff. Even the translations workflow will probably move to GitHub eventually. As such, they're probably going to need a GitHub account at some point anyway; it's only a matter of time.
nvdaes commented 6 years ago

Thanks Jamie, probably you are right. What you're saying is reasonable. Anyway, if someone don't know how to manage repos, at least we could ask the author to create a repo for the reviewed add-on if we have time. I think repos are important to review diffs.

derekriemer commented 6 years ago

@nvdaes commented on Jun 7, 2018, 3:31 PM PDT:

Thanks Jamie, probably you are right. What you're saying is reasonable. Anyway, if someone don't know how to manage repos, at least we could ask the author to create a repo for the reviewed add-on if we have time.

I don't think requiring a repo is necessary, just a github account. We only want to use an account so we can make the comments, and track the users review. I strongly oppose even offering to create a repo for the user. Maintaining the users addon is simply unscalable.

nvdaes commented 6 years ago

I think that a repo should be required. How can review diffs of add-ons without a repo? I think it's also impossible to review add-ons completely each time a new version is released. Cheers

derekriemer commented 6 years ago

This is putting too much strain on developers by forcing them to use git. We can keep a repo of previous binaries and diff off of them manually.

nvdaes commented 6 years ago

OK, for me flexibility is great, and giving all people chance for submitting add-ons. Anyway, if we maintain repos to see diffs, I think it's better we post them, so reviewers can benefit from the created repo and they have less work to do, since this avoids each reviewer needs to create his or her repo, not duplicating work and keeping it published.

derekriemer commented 6 years ago

We can't publish them on behalf of users without having push access to their repo, and there may be legitimate copyright concerns about doing this on behalf of an author. If individual reviewers would like to do this for authors and the author approves, I don't have a concern, but I don't think maintaining a repository should be a reviewers requirement, nor will I spend my time maintaining other peoples code. I also have concerns with authors who do not want a repo for whatever reason having the addons community making them feel obligated to have an addon placed into a repo. However I'd request that this not occur on the addons organization, so the addon team remains only a host for core infrastructure.

nvdaes commented 6 years ago

I thinkthat authors may not want their code hosted on a repo, and it's OK. But personally I think I won't spend my time making local repos to review diffs, since I prefer to get help from other reviewers. I don't think it can be problems with copyrights, since GPL 2.0 implies source code distribution in a free way. But obviously authors need to be respected and if someone doesn't want a public repo, it's OK. About nvdaaddons organization, when Mesar was with us, we maintained our code in the add-on team. OK, I confess I'm a Mesar's fan :) In our current organization, I removed my add-ons, for instance, but there are add-ons placed there, and I think we should let authors to do it if they feel more comfortable.

josephsl commented 6 years ago

Hi, I think a suitable compromise might be to word the review process text to say that having repo is encouraged/recommended. That way those who don’t want to use version control systems/repos can specify “N/A” when asked for repo links. Thanks.

From: nvdaes notifications@github.com Sent: Sunday, June 10, 2018 11:58 AM To: nvdaaddons/reviews reviews@noreply.github.com Cc: Joseph Lee joseph.lee22590@gmail.com; Mention mention@noreply.github.com Subject: Re: [nvdaaddons/reviews] Design form for submitting reviews (#2)

I thinkthat authors may not want their code hosted on a repo, and it's OK. But personally I think I won't spend my time making local repos to review diffs, since I prefer to get help from other reviewers. I don't think it can be problems with copyrights, since GPL 2.0 implies source code distribution in a free way. But obviously authors need to be respected and if someone doesn't want a public repo, it's OK. About nvdaaddons organization, when Mesar was with us, we maintained our code in the add-on team. OK, I confess I'm a Mesar's fan :) In our current organization, I removed my add-ons, for instance, but there are add-ons placed there, and I think we should let authors to do it if they feel more comfortable.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/nvdaaddons/reviews/issues/2#issuecomment-396072776 , or mute the thread https://github.com/notifications/unsubscribe-auth/AHgLkFJ8sYYqX9gVm8yl3QLJA6QOBgh2ks5t7WwsgaJpZM4Ue21Y .

jcsteh commented 6 years ago

I think we could certainly have wording around it being encouraged to have repos, as Joseph suggests, but I agree enforcing this might be a step too far.

As for having add-on repos under the team, while I understand Mesar's reasons for doing this, there are significant risks involved. It only takes one poor decision (whether malicious or by accident) to cause a whole lot of pain, pain which we might not even know about for quite some time and which may have wider consequences down the line. Mitigating risk is absolutely a good thing to do.

nvdaes commented 6 years ago

OK, we should encourage to maintain public repos without enforcing it. About risks, when we use the NVDA's translation system to translate add-ons, we need the add-on-team Bitbucket repo. Sometimes we have to create extra repos on Bitbucket for people who want to use the translation system, and it's difficult at least for me, not very comfortable, since GitHub seems more accessible or at least more comfortable for me. I'm using the add-on team uploading code of add-ons maintained by me, trying to do this just when code has been tested and considered stable, but if it's not OK, I understand. Just to report it. For now I have two URLs to push to Bitbucket team and my GitHub personal account at the same time with one command. If the translation system is moved to GitHub and add-ons can benefit, >I suppose that the organization should be use... This needs to be discussed. Also, I think that new versions have risks since in fact generally can't be reviewed. Though I think reviewers, when they have reviewed quite code of an author, could say that this person can be considered confident. Ideally all should be reviewed, but this is not possible and needs discussion. Maybe other communities don't review new versions of add-ons. I'm thinking on a Firefox add-on that was accessible and becomes inaccesible (I remember this especially since the author requested donations to release the new version and previously it was an interesting add-on, named ePUB Reader; I don't know if now it's or not accessible since I uninstalled it, just an example). Thanks.