PackageManager Download

[Dr. Fred]

There shouldn't be any grouping of the applications at all.

Why not ?

[Stead]

I for one really like crappish's ui, it seems clean, super user friendly, i'm not a big fan at all of the design that has been chosen so far, it seems non-user friendly to me, felt like i was using linux a bit to me, but thats just me!

i only hope someone tries to develop this gui to at least see how people respond to it

besides you could always have an advanced user button to have the treeview mode, theres no reason to why u can't still keep the treeview on the left, and use that nice description/list of stuff pane on the right

if the whole thing is in a dll as Dr Fred said, then i see no reason why any talented gui programmer, or anyone who liked crappish's idea, and is a good programmer, or even a bad one that can do it shouldn't give it a go

[Elledan]

Very professional looking, crappish *smacks Jaix with a 10-pound Unix-manual*

As for the HTML Control part, since Windows includes it by default, and many applications rely on its presence, why wouldn't ROS have it included by default either?

[Crappish]

There shouldn't be any grouping of the applications at all.

Nice arguments you got there. Let me point some counter-arguments to you. I presume you want everything to be on a single, big, list? Do you think that user A, who doesn't have wide knowledge about specific applications, and their function, wants to browse through whole list while trying to find an application for spesific purpose? He might not be have any clue which software does what he needs but he has a clue what he wants it to do. Hence if there are specific groups, and definately subgroups (my bad, forgot 'em on the first run), the user A can narrow the list of software to specific tasks, like instant messaging. From there the user can easily browse through few applications and find what he needs. Easily and without knowing the specific software he needs. Anyways, don't even know if you mean what you say.


To the rest of the gang. Thanks for the kind words. Like I said, I would like to see the UI in action but I can't force Dr. Fred to act according my whims (Oh, how would I love to force people to do stuff... ) and if he so decides he can freely keep the UI he has. After all, it is not the worst UI I have seen. And I completely understand if the coding of UI isn't amongst his personal favourites. For such occasions the ReactOS would need completely separate "department" which would eventually be in charge (have the last word) of ALL UI designs. (At least all of the stuff that is shipping with ROS) This is the only way to achieve good UI. And the department should be there quite soon as the project grows so does different opinions of UI design and usability. The lack of such quarter leads to lack of consistency and standards, which fundamentally is one of the basic flaws in i.e. linux. And as the ReactOS is, after all, aiming to be "another windows but better" the matter of UI should be taken into serious consideration, as it is one of the biggest advantages of Windows.

I just hope that open source scene would wake up and realize the fundamental part that UI plays in the game.

The rant is not aimed to anyone specifically, just to the general direction of open source community/development.

[Harteex]

My opinions..

The last picture by crappish is great imo, and it also solves the problem with sub-categorys.

I think it would be a nice idea to integrate the moz-control into ReactOS. If that wouldn't happen, just include it in the packet manager installer and let users choose to install it or not (if it's already installed or not that is).

I don't see how it would be too much work.. crappish has already designed this nice UI, so that means if Dr. Fred could use it, he could concentrate on the actual coding.

I also agree with crappish about the open source community and UI. It's a big problem.


so... Dr. Fred, please consider using this UI

[mjs]

The last picture by crapphis looks great but IMHO he should start with a tree view first (no need to implement it first and almost the same functionality).

Regards,
Mark

[user222]

There shouldn't be any grouping of the applications at all.

Why not ?

Most installers would not make use of the grouping system, so there might be a lot of applications that would not be grouped at all.

[rastilin]

I definitely prefer the latest mockup best, the design is excellent (the one with subgroups under the categories). What worries me is that we'll overload the servers if this thing really takes off, so it might be best to set it up with the bittorrent core.

In response to Mr Fred I should point out that, as this program is being coded from the ground up, there is no reason that any imaginable feature can not be implemented. We may have to have an Internet Explorer extension eventally anyway as several programs depend on OS HTML rendering to disply their interface.

[MadRat]

True, BitTorrent would be the user's way of repaying the team for their hard work and dedication. I like the idea. Perhaps the team could allow people to do that and allow for it to be turned on and off at will, too. Too cool of an idea.

[Dr. Fred]

There is no need for Bit Torrent the programs are download from the project's page (mostly sf.net with opensource projects).

In response to Mr Fred I should point out that, as this program is being coded from the ground up, there is no reason that any imaginable feature can not be implemented. We may have to have an Internet Explorer extension eventally anyway as several programs depend on OS HTML rendering to disply their interface.

We have an IE Control and that's the Gekko engien. But it's not part of the standart installation. I.E. you need the package manager to download it.

[Crappish]

What worries me is that we'll overload the servers if this thing really takes off, so it might be best to set it up with the bittorrent core.

Erm. I don't see any problem. Why can't the software just offer links to original sites? Straight to the installer. When you start the software it queries the master server about the list of available software but when the user downloads/installs i.e. MSN Messenger with it, it downloads the packet straight from MSN's server.

+ someone has to do a small application which on regular basis checks the master resource list, goes through it and removes all of the dead links.
The resource list could be PostgreSQL/MySQL database and the 'dead links' application would just set resource status as "dead" while the packagemanager would query all "live" resources. This way the actual links aren't lost and by fixing the URL it can be used again without typing all the content again.

[rastilin]

Erm. I don't see any problem. Why can't the software just offer links to original sites? Straight to the installer. When you start the software it queries the master server about the list of available software but when the user downloads/installs i.e. MSN Messenger with it, it downloads the packet straight from MSN's server.

So inefficient. That would use much more bandwith than is actually necessary. My reasoning is that just because the resources are availiable doesn't mean that we have to use them right now.

Having the requisite programs hosted on the www.reactos.com server means that we never inconvenience any other sites. Having the package manager be bittorrent based means that the hit to our site's bandwith will not be an issue even with massive numbers of downloaders. Anyone who's tried to download Windows 2003 service pack 1 in the last couple of days will understand what I'm talking about. This implementation will also give more control over the links (eg no "dead links") querying all the sites when logging on is also something I would like to avoid.

In essence, ReactOS is our problem, let's not make it theirs.

[Crappish]

Yes but the seeding is a problem. You really cant force people to seed the packets they've just downloaded and that's why I have my doubts about the bittorrent based solution, since bittorrent needs lots of seeders in order to work properly. If we have i.e. 100 applications in our selection then we need at least 10 seeders for each application. Of course many seeders can seed many applications. In any case, the thing here is that we need enough people who are willing to seed and have enough bandwith to share and I have my doubts about that. Although, I have been quite a short period in the community so I might be the wrong person to estimate this.

[Dr. Fred]

In essence, ReactOS is our problem, let's not make it theirs.

First I think that crappish is compleatly right with his uploader point. But there is another point: The package manager would not create trafic for the application developers. People would download the program in any case. If they do it manualy or with the package manager doesn't matter.

Most installers would not make use of the grouping system, so there might be a lot of applications that would not be grouped at all.

Sorry ? Either I don't understand your answer or you don't understand the concept of a package manager.

[rastilin]

Yes but the seeding is a problem. You really cant force people to seed the packets they've just downloaded and that's why I have my doubts about the bittorrent based solution, since bittorrent needs lots of seeders in order to work properly.

Several people will make an effort but I'm not planning to depend on that. The package manager should seed whenever it is running from copies stored in a separate folder, that way will require minimum deviation from the currenly proposed GUI and will be transparent from the users. Anyone wishing to seed need only start up the package manager and leave it running.

People would download the program in any case. If they do it manualy or with the package manager doesn't matter.

Not necessarily, I haven't heard about many of the programs on the list and seeing them lined up tempts me to download them as it would other people. With bittorrent we are minimizing our effect upon the other servers which I believe to be greatly beneficial. Sure it might not be a massive benefit for *us* *now* but it would prove beneficial to them by lifting some bandwith and massively so if ReactOS becomes popular. Imagine uploading updates to 5% market share, what about 10%. If nothing else, think of it as future proofing.