Heiko Tietze wrote:Sounds good. Are people who use a different download manager covered by "No, but I'd consider..."?

Hm, good point. It probably isn't completely covered. So I guess we should also add "No, I'm using a different download manager which I'm happy with (please state in the comments which one you use)"

Hey!

Sorry for being late again, having some busy days!
I clicked through the survey for a bit and it looks great to me. I'm not an expert for this though, so as long you (Heiko & colomar) are happy with it, I'd propose to go on!

So where to post this proposal? On my blog or is there an official "KDE Usability" one?

Lukas

boom1992 wrote:So where to post this proposal? On my blog or is there an official "KDE Usability" one?

I'd like to post on User Prompt (linked at Planet KDE). But feel free to retweet where ever you want. But first I need to start a clean new survey (e.g. I forgot to add a dummy persona when no one fits).

colomar wrote:So I guess we should also add "No, I'm using a different download manager which I'm happy with (please state in the comments which one you use)"

And that makes the starting question too complex. We better explain what Yes/No means. Most people here are smart enough

Heiko Tietze"
[quote="colomar wrote:So I guess we should also add "No, I'm using a different download manager which I'm happy with (please state in the comments which one you use)"

And that makes the starting question too complex. We better explain what Yes/No means. Most people here are smart enough [/quote]

Okay, works for me.

Heiko Tietze wrote:

boom1992 wrote:So where to post this proposal? On my blog or is there an official "KDE Usability" one?

I'd like to post on User Prompt (linked at Planet KDE). But feel free to retweet where ever you want. But first I need to start a clean new survey (e.g. I forgot to add a dummy persona when no one fits).

colomar wrote:So I guess we should also add "No, I'm using a different download manager which I'm happy with (please state in the comments which one you use)"

And that makes the starting question too complex. We better explain what Yes/No means. Most people here are smart enough

Perfect as long as it gets to the Planet!

Lukas

It tooks some time but finally there is the survey: http://user-prompt.com/please-participa ... ager-kget/
I'm looking forward the results especially the Kano method, which I didn't run online until now.

Heiko Tietze wrote:It tooks some time but finally there is the survey: http://user-prompt.com/please-participa ... ager-kget/
I'm looking forward the results especially the Kano method, which I didn't run online until now.

Hey!

Do we have any results of this yet?

Lukas

Two articles have been prepared (descriptives about current program and requirements for the next level) and are in the final review process now. Hope to publish them tomorrow.

What people think about KGet http://user-prompt.com/what-people-think-about-kget/
Using the Kano method to prioritize requirements http://user-prompt.com/using-the-kano-m ... uirements/

Heiko Tietze wrote:What people think about KGet http://user-prompt.com/what-people-think-about-kget/
Using the Kano method to prioritize requirements http://user-prompt.com/using-the-kano-m ... uirements/

First of all, thanks a lot for the study.

Having read over it twice now, with a couple days in between, I'm still unsure what direction to draw from it.
To me, there seems to be a lack of "Attractive" features to gain new users, while a lot of features we have lack discoverability.

The other thing what strikes me is the "medium" satisfaction with something like HTTP/FTP downloads, which belongs to the core functionality of KGet. That probably means that it is not too reliable and we have to work more on stability.

What do you think? Any more ideas?

Great job on the two articles. I actually just went through the second one just now. It was informative and I feel that I learned something on the Kano method because of it; thanks! I guess now we can decide how main content, sidebars, settings, toolbars, etc. can be done now!

I'm surprised by how bad Groups did. I guess it's time for new origination mechanisms that (1) can leverage tagging support, (2)mime type information, and (3) optionally use KDE Activities through the KActivites Framework. I see that last one as a stretch goal

I'm also surprised by how well prioritization did. Definitely, needs a prominent space in the UI.

Is multiple destinations just using symlinks? Or is it copying the same data to two different locations? Can Baloo be leveraged for this sort of task by using tags (kind of how Gmail places the same email in two different "folders" by using tags)?

Also, I can see a need for changing different settings. Such as using the notification center by default!

Like Lucas, I am also surprised that KGet does not have that many killer features. Though, I guess it time to work on presentation of what features are already there.

@Boom
From looking at the second article, the position of HTTP/FTP support on that scatter plot makes sense. You only get some satisfaction out of it because it is so required (kind of like a user saying "Well, how could you not support this feature?". I think the feature is stable but you don't give users much satisfaction by giving them "basic" features (you just avoid the risk of dissatisfying them).

_______________________________________
edit: My proposal

OK, I looked over the Heiko's articles and I think I have a few ideas where to go. Also, I want feedback on removing the tree view. So, everything is not completely fleshed out yet.

Here is an album of where KGet could go from here: https://imgur.com/UiDMtPr,FfmPzUW,NrOFv ... ,dH5zXqN#0

Content
It seems that with the new UI paradigms of Breeze, Tree Views are out of style. However, within KGet, a Tree View implemented the following features:
(1) Information of the download (a column per type of information)
(2) KGet's sorting mechanism (clicking on a column header would sort by that column's type of info)

Benefits for not using a Tree View includes:
(1) A better potential for following the HIG
(2) Info would not be cut off because a certain column did not get enough space.

I propose a detailed list view (somewhat inspired by Alex's new music player mockups [1]). Instead of a progress bar taking up precious space inside of each list item, I believe the progress bar could be either the top or bottom edge of each list item. I also feel there should be space between each item to have some clean space in the list.

However, that leads to what should go inside each list item.

Based on the analysis from Heiko's survey [2], this is what people want to see from a download (in order from greatest to least):

Code: Select all

Name of file
Download Progress
Remaining Time
Download Rate
Status
File Size
Destination
Origin/Source
Total time of Completion
Time paused/stopped


Obviously, KGet cannot show all of this information at once. Thus, I suggest KGet shows the following all of the time:

Code: Select all

Name of file
Download Progress
Remaining Time
Download Rate
Status
File Size


While, the status can be inferred through making the download look “disabled” if it is paused or stopped. This is sort of how options look disabled. The rest of this info (if it is deemed necessary) can be shown when expanding each download (to reveal the “details pane”)

I also think the download progress and the three subinfo (time, size, and rate) will have to be dynamic. I also think that the should remaining time will eventually become the time stamp. The file size should report something like “2/5 mb” until the download is completely finished.

It also seems that file name and progress are the two main important things. That's why I believe that the file name should get its own line (which also makes sure the file name does not get cut off) and the progress bar should be easily visible.
The other info can go into the details pane. It appears when the user clicks on the down arrow that is on each item in the list view. I have two separate design concepts for the details pane. (1) A simple dark box that contains the extra info and (2) a tabbed design.

The simple design includes things like the source, destination, and tags. It also has the ability to create new tags from the detail pane. There is also a button to reach the “Transfer Settings” dialog that KGet has (since verification and sanity checks scored pretty well).

For the most part, both designs include the same information. However, whereas the simple design as a button to create a dialog that represents the “Transfer Settings” dialog; the second design includes a tabbed design. So, the purpose of the tabbed design is to remove the dialog. The first tab includes the same content that went into the “simpler” design. However, the button has been removed. Instead, the second tab contains the files. A file can be renamed by double clicking it. Clicking on the mirrors button, leads to a third tab appearing to look and add to the available mirrors. Likewise, clicking on the other button allows users to look at the signature of the file and sanity check it.


Main Funtionality/Main Window Design
Also, revealed through the survey was how well users are familiar with the features of KGet. The feature that seems to be used the most regularly is “Pausing”. Other well known features includes Plasma integration, removing a download along with deleting it's data and adding a download through Kget's interface. However, deleting data through KGet's interface does not seem to be frequently done.

So, the most important actions (not counting button-less things such as drag and drop) would be:

Code: Select all

Pause/Resume
Restart/Stop
Integrity Checks
Prioritization


Also, I'm not entirely persuaded that people do not know about Metalinks is because KGet did not present it correctly. I don't think Metalinks caught on, even though they are a great idea. Not only that but supporting Metalinks would not cause great satisfaction but it's removal would not cause much dissatisfaction either [3]. Thus, I think creating metalinks should be included but not given any of the prestigious space.

Importing/exporting is usually a one time thing. So, that kind of stuff shouldn't be using the space needed for more important actions.

If users want a way to sort downloads, a sidebar could be added. This lets users sort even without the tree view. Once click on a category can sort in an increasing manner, while another click can sort the downloads in the reverse direction. However, in order to make sure the sorting sidebar works, some rules would have to be made (like a Finished download has 0 seconds left).

Settings
Groups don't seem to be scoring so high; but I think it can be recycled to a tag based system. For each tag, the user can add a default destination that kicks in if that tag is considered the file's “main tag”. The user can declare the main tag of a file from the “New Download” function.

KGet has several features to integrate with the desktop. There is the drop target, Notification Center Integration (which should be enabled by default according to the survey), and Clipboard integration, and a system tray icon (I thought these would die one day due to the task bar features Unity/icon-tasks copied from Windows 7). These could be grouped together in the settings.

I think it would be nicer if the task icon can become the “drop target” versus that weird icon that gets in the way sometimes. I'm not sure how possible that is...

KGet also includes network management features. However, monthly caps is best left to Solid or something even lower level (like NetworkManager). I also think the Web interface can just be added to the Network features.

All that was left out of the Advanced Settings was the configure on start option, the the option of what to do when a file already existed. So, I just added a global “action on completion” feature (which defaults to “do nothing”) just to fill some of the space. Though, I don't know if a global option makes sense for that.

[1] viewtopic.php?f=285&t=122273
[2] http://user-prompt.com/what-people-think-about-kget/
[3] http://user-prompt.com/using-the-kano-m ... uirements/

boom1992 wrote:To me, there seems to be a lack of "Attractive" features to gain new users, while a lot of features we have lack discoverability.

Absolutely. On the other hand, if we ask users whether or not they can live without plasmoids for instance, we would get a high rate on indifference too. But not if we ask them about 'flexible desktop integration' or the like.
I tried KGet over the last couple of days via FlashGot plugin on Firefox. To me it adds more distraction than solving problem; but I'm lucky to have high-speed access. So the question is who is your target user. The guys who just wants to get notified (with the goody to tweak the background process seldom) or those people who need it to manage the process because they live in the rural area. For the first one it would be a simple plasmoid, second goes in the direction that CTown suggests.
About the proposal: It shows nicely how to present relevant stuff and hide the less important information. What I'm missing is the workflow from clicking a link to accessing the resource. I will try to create a mockup tomorrow.

Is this what you meant by the workflow?: https://i.imgur.com/n3Y1dql.png

Background
People do not really use filesystems. Most just don't seem to get the concept of a file hierarchy. Microsoft tried to enforce libraries for Windows 7 (by that a special folder for music, videos, etc.) but I'm not sure how that worked out. However, I think the fact that OSes usually come with these libraries is a great starting point for organization. Though, on Linux, we could never enforce such rules. To give an example, my brother does a lot of work in Photoshop. However, he saves does files all into one file (I guess he never imagined how quickly the number of files will grow). So, to look for files, he uses the search mechanism of Finder (the Mac OSX file manager) and it works really well for him. However, would that work out if he was never the one who named the file himself? After all, no one really renames a file before downloading it.

Matt
Matt finds a couple of potentially nice PDF files and a few nice pictures for his geology project. Thus, he uses KGet to download it to the file so it goes to the correct location for easy retrieval later. However, resources like these have tend to have arbitrary file names. As, Matt hates searching for content, he uses KGet to tag his files derived from the abstract of these third party reports and a few descriptive words of the pictures. Matt can also drag players onto the UI to start their downloads to easily tag them.

Also, Matt can add Metalink creation to the toolbar (since it was a feature he seems to likes according to KGet's Kano analysis). However, the usual way to use a torrent or metalink file in KGet is to actually "open" the torrent or metalink using KGet. It would be nice if there was a more intuitive way to do this (a complaint of GreatEmerald).

Susan
Susan consumes a lot of media. Thus, she uses Web browser integration to download media with KGet. She sets up all downloads to save to the right location by mime types. This way her favorite applications can always find her media without searching her entire Home folder.

However, it would be kind of cool if there was functionality to limit KGet's bandwidth during a video call. By I think that is out of KGet's scope. The most I can think of is pausing all the downloads or KGet notices a consistant slowdown of bandwidth and just pauses downloads by itself or limits its own bandwidth use.

Berna
I doubt KGet could be a good tool for syncing content. There would be just so many better tools to get the job done. Also, the only uploading KGet really does is torrents. Sure, she could prioritize downloads and keep a log of everything she does (including the tags she placed) to import/export but that is all of the content retrieval she is likely to get out of KGet. I guess OwnCloud could be used so that it (1) asks KGet to export the transfer list, (2) gives an instance of KGet on a different computer a new list to import, and (3) makes the other KGet download those files from the original sources. I guess this would avoid the need for uploading data (which is usually multitudes slower than downloading) but this doesn't seem reliable as one is relying on third-party sources for a backup (which is the original in this case) of the file.

I think a document editor plus a service that can backup filesystem metadata (where Baloo stores tags) is better for her.

CTown wrote:Is this what you meant by the workflow?: https://i.imgur.com/n3Y1dql.png

Kind of, or rather: such an UML diagram would be nice. But I'm not an expert in modeling business processes. What I'm thinking about is this:

http://vo5jbp.axshare.com/kget_simple.html

Well, its only the half. The idea is (or was) to create two mockups, one simple in terms of reduced functionality and the other full-featured, both including interactions. I started with the simple mockup that is just a notification plasmoid with a few global options to manage the process. The idea addresses doubts if we really need to group or prioritize downloads with todays bandwith. There are rumors that monthly cap will be reintroduced at 50GB or so, but those settings should go to the network manager itself because KGet will not be able to see streamings, wget, or pacman/zypper etc. Actually Philip never has to wait for the resource he is downloading so it makes not much sense to organize it with KGet. What Philip needs is to limit the download in case he listens to audio streams, which could be an activity.

I wrote "was the idea" since it takes much time to create the mockup (with limited benefit). I would need another week for the advanced version, and haven't actually any good idea how to design KGet around groups (that's what I'm missing a little bit in your proposals). But your user stories are brilliant and perhaps I'm wrong with the simple version. What I don't understand is why metalinks and torrent should be relevant for any user. The program should download any type and protocol, or not?

Nevertheless I'd suggest to progress with small steps with both the full-featured and the simple version, and Lukas does the cherry picking later.

I started with a mockup and then I found out that KGet have the feature and also the workflow works really well.

+ download selected files from a page works fine,
+ save a single file works fine
+ the list view in the KGet GUI is easy to understand. You can add and remove items easy.
+ the icon bar is compact

the settings could be more compact but there are a lot (maybe to much) settings.

- the group function could be more visual for the users
- I don't like the detail view as it is now. would it be possible to have a panel ob buttom and when you select a download you get the detailed informations for this download file like in ktorrent (maybe not to complex but it is a good starting point)
- as CTown expected a notification integration would be more useful than the application icon in the status bar. maybe you can switch from the notification to the application.
- when I use rekonq the via kget downloaded files aren't shown in the rekonqueror download page.
- it would be nice to have a search function like in the mockup from CTown and also a navigation via day, server,... than you don't have to remove items from KGet to have an overview
- streaming support would be cool. When you look how much firefox extensions are available for streaming pages then I think there is a "need" of download streams.