boom1992 wrote:What do you think?

It does not thrills me like KTp's vision. Maybe because I don't see the bright future where you rule the world. But it's now on a similar level like the other example, that colomar posted. Rather descriptive than visionary.

Heiko Tietze wrote:It does not thrills me like KTp's vision. Maybe because I don't see the bright future where you rule the world. But it's now on a similar level like the other example, that colomar posted. Rather descriptive than visionary.

It seems like many maintainers shy a bit away from the grandeur that such truly "visionary" phrases transport. I'm glad I could convince the KTp team to include such a sentence

Well... If you guys have ideas on how to take over the world, let me hear them!

OK, I tried to make a new persona, based on a person who shouldn't be that hard to find. I guess it can be matched up to Susan, just replace "work". Actually, that would be better as Susan does do some bandwidth-intensive activities such as video chatting and uploading videos. Still, how often do we use multi-ethnic names (Aya is Japenese for design/color and Arabic for wisdom). Also, I think this can help us establish some requirements.

Aya is a media junkie and a neat freak. She uses download managers to make her multimedia life much easier. First, she uses browser integration to send a download to her download manager. She uses "Groups" to make sure the download gets to the right place (because using the mime type wasn't enough). Not only that but she can take a download and schedule a task around it. This way, her download manager looks for a variation around the URL in case a newer download is found (the download manager is given the pattern for the URLs that the multimedia uses and periodically checks if the next increment in the URL pattern is found). Finally, she makes sure downloads are paused during a certain period where she regular uses her computer to get more important work done.

Also, I have been trying to read on why people use download managers. I found a LifeHacker article on it, here is my summary of some comments that caught my attention. I also think this website has good information.

Edit:

I just realized that the Etherpad says that the list I made is a set of requirements. I want to clarify that that I just made that list as the features in KGet that I know of in order to reorganize those features. But then Heiko and colomar said I wan't at a point where I could draft a feature set for a new version of KGet.

I also want to point out that while torrents are pretty important, I think torrent clients give WAY to much information by default. If I had to guess, I would say that the only the only piece of information from torrents that most people want (besides the same information from HTTP downloads) are the numbers of seeders to the torrent. But of course, if there no seeders, the user would be able to understand that when KGet sits there for a few hours doing nothing. This could be easily solved by KGet warning the user that the torrent has no source at the moment and letting the user choose what happens next. I honestly believe that KGet should be the first place people go to when the words "torrents and KDE" come to mind.

Thanks for all replies (and especially the summary of comments; I hate those infinite lists). I updated the pad accordingly:
* Vision by Lukas (I have no better idea at the moment)
* Scenario by CTown
* User Stories: moved to a separate paragraph (should get integrated into scenarios and requirements)
* Requirements: moved grouping stuff from management to an own epic

It seems to me that we have a pretty good understanding now what we are talking about. So I would suggest to have a hangout to discuss the next steps. For instance, I believe the transfer group feature has high costs (a lot of configuration, selection etc.) and is used rather rarely. We could focus the survey on those aspects and make the app either simple (remove requirements completely or hide them in the first place), and/or ask for further needs. In order to keep the number of questions tolerable we should discuss which function is fix and which are debatable. And this is done at best in a direct talk, IMHO.

Okay, since yesterday I have been thinking a bit about what people actually use download managers for. I think a lot of the reasons why I used them (or in this case used KGet) is that there was limited bandwidth per connection to a server (solved with multi-threaded downloads) and limited bandwidth on my side of the connection (not really solveable, but having control over which download should go first etc partly solved it). Both of these cases don't really make sense anymore, given that connection got much better in recent years.

I also talked with Aleix (Pol) about the idea of organizing the Download folder. As far as I understood him, it is not a problem, as finding files on disk is a problem that is rather solved by baloo and meta-information.

Another point that was made is having an overview of how much bandwidth was already used in a given timeframe. This would better be solved on a system level (as in android), given that many applications take bandwidth and not only the download manager.

So the only point I am left with right now is integration of downloads into KDE (that means HTTP, FTP, Torrents), which is currently done very poorly. I think having KGet just do this is a bit overkill, it can be done much easier with just plugins for different browsers (register the transfers in the notification area, like KGet can already do) and KTorrent (doing the same with notifications).

Can someone give me perspective on this? Do you agree?

Lukas

boom1992 wrote:Can someone give me perspective on this? Do you agree?

I'm not sure what you are talking about. To abandon the project?

Killer features are IMHO:
* bandwidth optimization (as you said plus resuming)
* download of many items from one site (could be achieved by a plugin, but then browser specific)
* easy integration into KDE (just drag an item onto the KGet plasmoid)
* full integration into KDE (global notifications)

* grouping and auto assignment (separated from the rest because it's not the primary feature for all users)

Heiko Tietze wrote:

boom1992 wrote:Can someone give me perspective on this? Do you agree?

I'm not sure what you are talking about. To abandon the project?

Killer features are IMHO:
* bandwidth optimization (as you said plus resuming)
* download of many items from one site (could be achieved by a plugin, but then browser specific)
* easy integration into KDE (just drag an item onto the KGet plasmoid)
* full integration into KDE (global notifications)

* grouping and auto assignment (separated from the rest because it's not the primary feature for all users)

Hey!

Well my point is that if there are about no valid use cases, I do not see why anyone would use it in real life.
I'm just trying to get a discussion on this going, that doesn't mean I don't want to continue here. I just see it as another possibility that this analysis could lead to.

Bandwidth optimization => I'm not sure if that is still valid, as I said above.

For KDE integration, it's probably a bit overkill to use KGet. After all it is only a small part of this program.

Lukas

A lot of people still use Download Managers, for OCHs that is. Look at JDownloader or pyLoad.
But that might not be the right way to go for KGet, way too shady in that corner...

I use a download manager because I used to close Firefox before a download would finish or perhaps close the lid of my laptop and Firefox would cancel the download. Though, I think at this point Firefox is getting better at resuming. Not only can Firefox pause and resume now (even though it doesn't seem like a stable feature on Firefox) every download manager in the world supports this.

However, I think "Groups" is the killer feature for me. If I am working on a project (like for school for instance), I can go get random pdfs with examples, data sheets, pull lectures/practice tests from other college campuses into the right folder right away using Groups. You guys are going to make me laugh for days by taking out my favorite feature! Though, it wouldn't be the end of the world. It's not that many files to begin with. Though, something more powerful could probably be done using Baloo (as Alex said) or tags. People seem to enjoy tags over folders anyway. For example, in Gmail one can give an email multiple tags and never think about which folder the email is in. Though, I do understand that most users don't create/use that many folders.

@boom1992
I think those features make sense. Not many Internet connections can handle multiple downloads, video chatting, and most importantly Netflix (the king of Internet data usage). It would kind of be cool if KGet noticed if some of these activities going on (like when the downloads slow down by a lot) and then KGet can further limit the bandwidth of the downloads to ensure a great user experience on Plasma.

I don't think the integration features might be "killer" but it would make KGet a lot simpler if each job (mass downloads should count as one job) showed up in Plasma. That way KGet's main interface won't have to be opened so much (downloads should be a part of the work flow - not the work flow itself). Plus, the "main" actions could easily show up in the notification center. This is close to what Firefox does: (1) there is the quick downloads menu by clicking on the download menu (corresponds to KGet's Plasma integration) and (2) the download library (which corresponds to KGet's main interface).

@hirnsushi
Are one-click hosting sites still popular? Also, I think KGet does support these sites. Though, I think these sites usually update once and a while to break download managers.

CTown wrote:Are one-click hosting sites still popular? Also, I think KGet does support these sites. Though, I think these sites usually update once and a while to break download managers.

I don't have absolute numbers to support this, this is only based on the usage I see in the casual users around me, biased towards my age group and location. At least in this group torrent use has pretty much faded, because of security reasons, while the usage of one-click hosters is still pretty high.
KGet doesn't support this and I don't think it should, but it is a "valid" use case.

I agree that it should not support One-Click-Hosters. We had experimented a bit with it some years ago, but it did not make much sense.

@CTown: I'm not sure if managing bandwidth makes much sense anymore. At least for me the bandwidth has gone up rapidly the last years. When I started working on KGet having a download manager made absolute sense to me, I used it daily. Since moving to a Fiber-Connection and even now as I'm being downgraded to sth like 6000 MBit/s shared with 3 people, I have not really used it. Downloads are mostly fast enough.

The only thing that could make sense is sorting files for me, but even there seems to be no real concensus.

boom1992 wrote:@CTown: I'm not sure if managing bandwidth makes much sense anymore. At least for me the bandwidth has gone up rapidly the last years. When I started working on KGet having a download manager made absolute sense to me, I used it daily. Since moving to a Fiber-Connection and even now as I'm being downgraded to sth like 6000 MBit/s shared with 3 people, I have not really used it. Downloads are mostly fast enough.

It depends: With a few torrent transfers going on, for examples, bandwidth (especially upstream for ADSL) can still clog up pretty quickly.

And you need some features!

I also like the idea with the priority for web serfing, streaming pages, ... instad of a fast download and you can't watch a streaming page or send/recive your mails. That would be a nice feature.

Again, should we discuss this in a hangout? Otherwise I can prepare a survey based on the considerations at the pad.