Complete online database wipe (once every few years?)

[Tomorrow]

Since there are so much stuff in the online database and no-one is sorting and removing the nonworking ones i think a simpler solution would be to simply wipe the whole thing clean.

 

Besides over time lot of stuff will be re-shared.If few years seems too little then maybe once every 5 years?

[Omniferum]

I'd agree with a starting from scratch but periodic wipes seems a bit much. If floele split the online db into two types, admin/volunteer moderated and 'whoever wants to submit'. That way we can at least have a 'guaranteed working' category or at least if it is reported will end up getting fixed and a anon submission that can be periodically wiped yearly perhaps.

[Etz]

+1.

Also Would it be possible to add option, to delete apps from online DB?

Basically that every user can delete things that he has created and shares?

 

Also I pretty much like Omniferum idea.

Then we could housekeep always working app templates and clean database.

Edited July 26, 2011 by Etz

[crrodriguez]

I see in the database that there are > 1200 entries which have not been updated in more than a year... but though they are too old, Im not sure if that's enough criteria to delete them from the system.,

[Etz]

Hmm... maybe create a script, which deletes automatically from DB, if app is not renewed in X period of time?

Edited July 27, 2011 by Etz

[crrodriguez]

Hmm... maybe create a script, which deletes automatically from DB, if app is not renewed in X period of time?

 

The database can take care of that by itself, however what needs to be defined is X in period of time and what other conditions to use.

[Tomorrow]

No sure about automatic removal after x amount of time if not updated.

Some things just don't change very much over the years like FileHippo and such.

 

Still think that special case wipe once every x years for all things would be better.

[Etz]

I guess, at the moment, 1 time single wipeout would be enough...

Theres too many, not working and duplicate templates.

 

Also I wonder, if I wont edit my shared template, will it re-upload itself automatically, if it doesnt exist in DB,

or I have to open it in edit mode and hit OK for that?

[Omniferum]

No sure about automatic removal after x amount of time if not updated.

Some things just don't change very much over the years like FileHippo and such.

 

Still think that special case wipe once every x years for all things would be better.

 

Yeah but those that are admin approved/working won't be auto-wiped, assuming my suggestion is implemented.

 

I still think an auto-wipe for anonymous submissions should be in place.

[floele]

Best would be to set up a web interface for cleaning or rating apps, but that's not done quickly. I've been thinking about that for quite a while already.

[Omniferum]

Actually on the note of auto-wiping I think it would be a good idea to include the ketarin version used to develop the template. So if it posted to the online .db you get "Ketarin v1.6" or whatever was used to do it.

 

It feels like the .db has never been wiped so has a lot of old Ketarin templates whilst the newer ones would have better functions implemented.

 

Future implementations aside I still feel like just a straight wipe now and starting from scratch would be a good thing.

[shawn]

I think it ought to be better coupled to an author. Probably require membership in the forum to be able to submit to the online db. Then, the script on the AUTHORs machine has to be processed successfully at least once per year in order to remain in the online db. The author then enters their login and a key provided by their profile on the forum to enable online database interaction.

 

WordPress just announced that they were going to start "sunsetting" plugins and themes that weren't updated within a certain period (I think they said 2 years). In any case, Ketarin wouldn't be the first to do it.

[necrox]

I think it ought to be better coupled to an author. Probably require membership in the forum to be able to submit to the online db. Then, the script on the AUTHORs machine has to be processed successfully at least once per year in order to remain in the online db. The author then enters their login and a key provided by their profile on the forum to enable online database interaction.

 

WordPress just announced that they were going to start "sunsetting" plugins and themes that weren't updated within a certain period (I think they said 2 years). In any case, Ketarin wouldn't be the first to do it.

 

*bump*

Yes, please! Would be a nice feature. If you have questions regarding a specific daabase entry you know at least the author's name. ^^

[Guest LeBaux]

Adobe Reader (en) The downloaded file is not a binary file type (text/html; charset=UTF-8). Possibly there is an error page. Status code: 200 (OK) (http://www.filehippo.com/download/file/fe877141894be605206576a6c0667ce003247989570452117dfbb9b2d234fd54)
Adobe Flash Player IE The downloaded file is not a binary file type (text/html). Possibly there is an error page. Status code: 200 (OK) (http://www.filehippo.com/download/file/191c511508b19bd6d4b66a58d67f381dc3107bb9be4b854596c0dd704faa3108)
Adobe Flash Player Firefox The downloaded file is not a binary file type (text/html). Possibly there is an error page. Status code: 200 (OK) (http://www.filehippo.com/download/file/75a88a50992e959947dd804bbc17546a26f912e9d1c21a522420cc85d87ca339)
Firefox Win-US [official release from Mozilla website] The remote server returned an error: (404) Not Found. (http://download.mozilla.org/?product=firefox-{version}&os=win〈=en-US)
GOM Player The operation has timed out (http://www.filehippo.com/download/file/4c5eb2af651aa964511e72d2b2cf467dbe3ba3a7ba135f817558919322fec33b)
ImgBurn The operation has timed out (http://www.filehippo.com/download/file/280b3e8253bfb3baf746b70f0ea22c4029bbbda979830ada79b8cc7116cc6030)
Google Chrome The operation has timed out (http://www.filehippo.com/download/file/0d5524ca882e7ee14a5b6d0496504d0586158d61f354ddac5127c4d0e2a38b7a)
PDF-XChange Viewer The remote server returned an error: (404) Not Found. (http://www.tracker-software.com/download/PDFXVwer.zip)
Microsoft Security Essentials (32-bit) The operation has timed out
Opera The operation has timed out

 

this speaks for everything.

[shawn]

The filehippo downloads will have issues if you don't set them to download 'exclusive', most of the others have had major site changes over the last year.

[kerframil]

Warning: long post ahead ...

 

As far as I can tell, Ketarin exhibits a characteristic which makes it extremely difficult for anyone other than the author to assist in maintaining the database. Specifically, if I define an application and choose to share it, I will be able to make future changes to the package and these changes will be automatically propagated to the online database.

 

However, if I import an application that was originally defined by someone else then it seems to be impossible to make any changes. Ergo, the situation is that anonymous users cannot import and correct existing packages, even though anonymous users are trusted to share (upload) whatever they please in the first place.

 

In my opinion, this is bad because the only recourse is to either [a] upload a duplicate (albeit working) package, further cluttering the database contact whomsoever has the ability to modify or cull the detective package. In the latter case, it's not even clear who is doing the job. It's one thing to mark packages as "abandoned" but it's rather frustrating not to have the ability to import and subsequently fix anything. All of which leads to the online database accumulating yet more dead wood.

 

If a submitter makes the mistake of deleting the original packages from the database, the same issue arises. For example, I submitted the "Thunderbird ESR (tete009)" package. It was removed from my database at one point. I later added it with the intention of marking it as obsolete because tete009 has stopped producing ESR builds. Of course, I can import the package but any changes I make will no longer propagate. Hence, it is impossible for me to flag the package as obsolete!

 

In my most recent submissions, I have started entering my email address in the Notes field in the vague hope that, should someone discover an issue with one of my packages before I do, they might contact me. There has to be a better way ...

 

IMHO, before going so far as putting in the hard work of creating a web application - essentially constituting a brand new application - the existing framework could be improved. I have some ideas which I'd like to put forward:

• Move away from the notion of purely anonymous submissions in Ketarin and encourage the notion of maintainership (at the very least, an email address should be mandatory as a unique identifier) (*)
  
• Have the backend database support a specific field to indicate that the application is abandoned/broken/obsolete and a method for a user to flag an package as such - directly from within Ketarin. Any user should be allowed to do this but not anonymously.
  
• When a package is marked as such, ensure that the maintainer is notified. For instance, an automated cron job that sends an email. Further, the Ketarin client could alert the maintainer somehow.
  
• The maintainer should have the opportunity to correct the package within a given period of time e.g. 30 days. If the maintainer is responsive and does correct the package, the Ketarin client should allow him to update the package and remove the flag that indicates it as being abandoned or broken
  
• If the flag does not change within the grace period, the backend should automatically delete the package (again, a cron job).
  
• The maintainer should retain the right to remove his or her entry (which doesn't stop people from forking/re-submitting the package)
  

Automatic deletions may appear contentious but I don't think it would be a problem. For one thing, even if the maintainer fails to respond in the grace period, he/she would be free to re-submit the application at any later time from the local database Also, if the maintainer really has abandoned a package, any other user should be free to submit a newer/forked version. With the current system, this leads to duplication but this problem would be curtailed by automatic time-based deletion of abandoned packages.

 

Other situations that might arise would be eased by deprecating the currently anonymous nature of the database. For example, let's say that someone has already submitted a package that I have defined. Perhaps I consider my approach to be better (due to safer regexes, improved metadata, silent installation support or whatever). If the email address of the submitter is known then I can contact the author saying "hey, would you mind incorporating these improvements?". For this kind of thing, the idea of having a web interface starts to look attractive, allowing these discussions to occur in the open. Something along the lines of the AUR (Arch User Repository) might be a good fit.

 

Disclaimer: these are merely my ideas and do not constitute an expectation or demand that they should be implemented! I'm merely someone who finds Ketarin extremely useful and would love to see the quality of the online database improve.

 

* shawn's idea of tying to a forums account seems like a good one

[shawn]

I didn't read it all, sorry.

 

However, you can fix any application - you just have to change the guid and give it a unique name. The easiest way to change the guid is to select the row in Ketarin, CTRL+C to copy it, paste it into notepad, change the guid to something else (usually on line 3, must be lowercase and hexadecimal), select all, CTRL+C, return to Ketarin and paste it back in (CTRL+V). I've got this down to a science now - it takes me less than 3 seconds total. At this point you only need to make sure that the 'shared' box is checked and it has a unique name for the online database.

 

Anyway... I would really like the online db to show the submitter's handle within the online database. Right now you really never know what you're going to get. For the most part, I stopped collecting from the online database because some of them are just total crap.

[necrox]

As i said before. I love the idea of linked database entries with an email address or user name.

 

@shawn, I also stopped collection from the onine database. On the other hand I offer my scripts online for new /interested Ketarin users. In order to keep good entries up we should think about a rating system for "working" / "not working" entries. Or a database wipe once a year to delete old entries.

[shawn]

Me, too, necrox. All of mine are included, but it's rare that I'll find one that works the way I want it to in the db.

[kerframil]

However, you can fix any application - you just have to change the guid and give it a unique name. The easiest way to change the guid is to select the row in Ketarin, CTRL+C to copy it, paste it into notepad, change the guid to something else (usually on line 3, must be lowercase and hexadecimal), select all, CTRL+C, return to Ketarin and paste it back in (CTRL+V). I've got this down to a science now - it takes me less than 3 seconds total. At this point you only need to make sure that the 'shared' box is checked and it has a unique name for the online database.

 

Thanks for the tip. The problem is that this leads to further cruft and duplication because the requirement for a unique name means that it's impossible to overwrite the original (redundant) entry in the database. In this case, what I need to do is mark an existing application as obsolete or - better yet - delete it outright because it's no longer provided by upstream. Unfortunately, neither option is within my grasp.

 

As a side note, you can generate a truly random GUID by running the following in a Powershell prompt (use right-click to paste the command from the clipboard into the shell, left drag to select the resulting GUID, then right-click to copy the GUID to the clipboard):

 

[guid]::NewGuid().ToString()

[shawn]

Agreed, that is a problem. But you can always alert Flo to the broken profile here though - just gotta let us know.

 

You can also generate a unique guid with DQSD by typing in:

guidgen

[Guest mida]

whats about this, example i have a application profile that im shared, and every time if i get no errors on update check or Ketarin take a new update from this application, its send to online DB that this application profile work with actually date time (UTC or samething else)

naturally, if on a update check get a error that this file not longer available etc. it sends a error message to DB.

Admin can than delete application update profile in the DB, all with a this error, after x errors or else

 

[Tomorrow]

Glad to see this topic is still alive. As it should be. I agree with everything that kerframil said and he offers some great ideas in his post. For now when i abandon an app i do this:

 

1. Add [ABANDONED] string to app name, disable it and publish to online DB.

2. Remove Edit>Advanced>"Share this application online with others" checkmark.

3. Export this app with the abanoned string in xml name.

4. Include the abandoned xml in archive so if i decide to re-use it in the future i still have the original and i can import it and publish any updates to online DB.

 

Its fairly redundant keeping around old templates but you never know when you might need them.

 

EDIT:

 

Was wondering. Could two people who both have the same app down to the GUID - could they co-update an app?

Personally i dont see why not unless one of them already has an app with this name/GUID in their list.

[Guest chr0naut]

Why not make the flagging of non-functional online database entries a programmatic one; i.e: The users themselves do not have the authority to flag bad entries, but the Keterin client program does.

 

This does NOT remove the entries, it merely flags them as failing execution. It also does not remove the entries from the local client database, so if a database entry starts working again, the program can remove the non-working entry flag from the online database.

 

If after (say) 3 months the online database entry is still not working, then it is marked as permanently dead and is either purged or preferably, marked dead (and no longer visible to users). An update of the online database status is then flagged back to the client program saying that "This online database entry has been marked as dead after 3 months with no successful executions" and an option is available to the user, to remove the entry from the local database, as well.

 

This makes online database maintenance entirely automated, with no issues of trust and is the most cost effective option going forward.