[haiku-depot-web] Re: Introduction article

  • From: Andrew Lindesay <apl@xxxxxxxxxxxxxx>
  • To: haiku-depot-web@xxxxxxxxxxxxx
  • Date: Fri, 29 Aug 2014 23:57:35 +1200

Hi Stephan;

Yes, I agree, it is nearly time to do that which is great. :-)

One thing that I have been thinking about a bit for the last week is authentication. As there are already a number of systems in play already (TRAC, Drupal, git, ...?) the authentication is already fragmented. Adding another system and subsequent authentication credentials into the mix will make things a "little bit worse". Thinking long-term this may be confusing for people. Is it worth expending effort and delaying now in order to look at trying to harmonize authentication for these web / application-server systems going forward or is this too big a block of work to get into right now?

I do think a dev / test / staging deployment separate from the production deployment is necessary. As you mentioned, the 'current' deployment could (conveniently) become the production deployment and another environment could host the continued dev / test / staging deployment. All of the system's storage is in the single Postgres database instance so it should be easy to backup with pg_dump -- that should be happening on some schedule before we let people start loading material in. Oliver; you'll have some ideas on these sorts of things?

I think your approach to permissions makes sense. How can that be best managed; on a mailing list?

I see in the logs there are recently lots of API look-ups on packages that are not present;

org.haikuos.haikudepotserver.api1.support.ObjectNotFoundException: the entity Pkg was not able to be found with the identifier haiku_devel

I'm trying to figure out where those are coming from -- could that be the desktop application? Oliver; could I get access to the Apache2 logs to see if that helps me understand the source of those?

Regards;

On 29/08/14 10:39 pm, Stephan Aßmus wrote:
Hi all,

what is the general opinion about giving the web app, and the
installation in particular, more exposure in the project?

I synced up the HaikuDepot categories with the web app and then went to
the web app and defined categories for the first two pages of packages.
But there are already 17 pages... :-)

Then I wrote some sort of "call for help" message to haiku-development
(which I did not send). It contained many more paragraphs which
introduced the web application, and I realized 1) I should ask for
opinions here first and 2) it should be an article or blog-post on
haiku-os.org.

Basically, my original call for help with regards to the web app was
burried somewhat deeply in an unrelated thread. Although I received
several offers for help in private mail, the whole effort was unknown to
some (many?) in the project for quiet a while.

With the introduction article, I would like to describe the purpose and
status of the project and also point to the current installation. I know
that it was meant as a "test" install. But I don't know if or when the
status quo would even change. It seems to work really well as a "real"
installation to me. And even if it is replaced, I guess the data-base
would be migrated in order not to throw away all the work.

I would like to encourage people to create user accounts. I don't yet
have a clear idea how to proceed from there and pick users who should
get more permissions. I guess we should just have them apply and revoke
permissions if there is any abuse. "Real harm" can only be done with
translations (i.e. rants) or screenshots (nude pics...), not so much
with categories. The users with permissions to write translations could
perhaps be matched to the existing Pootle team.

Thoughts? I am really happy with how the web app turned out, I think it
can give a positive impulse and it is complete enough to provide an
opportunity for people to contribute and collaborate, just like
originally envisioned.

Best regards,
-Stephan




--
Andrew Lindesay

Other related posts: