[Iss Mneur]

This could be fairly easily setup with a proper version number system and some meta-data for the installer to use.

Indeed. I did not said that with the meaning "how could we do this", but rather as "there will be a little more work than just downloading one big file if it's different than what you have"

Of course, I didn't mean for it to sound like I was saying "this was the only way to do it", more a it is doable with some technical specifications/design so that the mods would be able to provide input to the process.  Because if this is going to work in an easy and not complicated manner like Hellzed wants we need the whole thing to be supported by a large number of mods.  As well I am sure we would not be the first ones to have this problem so there is likely existing solutions for this very problem and it would then be a task of just integrating.

@The E:
Hellzed proposed an installer/launcher for Linux, this spilled over to a new installer proposal. Somehow, if we manage to make a unified and platform-independent launcher/installer, I feel it might be worth building it as an one application.

Yes, I too, I think this would be of most value if it would work on the fs2_open supported platforms because the problems that Hellzed talks about apply to all platforms, though linux being what it is, the problem is likely more acute. Of course, this is probably the reason that this discussion has gone off on this angle to start with.

Me, I'm mostly a designer than a coder, though I wrote code too. Hellzed said he will try some mock ups, and I'm still waiting for them before putting anything on the table.

I too would like to see what Hellzed comes up with.

Iss Mneur, would you like to join this project as a coder, assuming we manage to truly start it?

I would, but as I stated in the my last post I really would not be able to dedicate any time to coding it until after October starts, but would be able to put some input into designing it.

[kkmic]

Alright, so we have a lot of parameters that we might need to pass to the program. I agree that for the moment, the registry option is the best we have, but I'll keep studying the problem.

Why is bad passing parameters via a text file? (technically speaking)
1. The launcher might not be able to write the file or to rewrite it's contents
2. FS2_Open might not have the right to read the file the launcher wrote

Other posibilities?

I would, but as I stated in the my last post I really would not be able to dedicate any time to coding it until after October starts, but would be able to put some input into designing it.

Good enough. I always said that many project fail because people who code them are not able to see the final shape of the product for one reason or another.

Design is law

[The E]

I think you're misunderstanding what I said. The Launcher needs to read and write to the registry if it is supposed to be compatible to all FSO builds.
Passing parameters via text file is necessary to a) avoid storing said parameters in the registry b) Keep them around in a human-readable format and c) because of the string length limits portej05 mentioned.
Now, the only scenario where the Launcher is not allowed to write this file to a location where FSO can read it (IIRC) is on Windows Vista and later with the Launcher not being in the same directory as the FSO exe. YAL just writes to ~/.fs2_open, and I imagine Soulstorm's Launcher does the same thing.

Design is law

No. Functionality is.

[kkmic]

I think you're misunderstanding what I said. The Launcher needs to read and write to the registry if it is supposed to be compatible to all FSO builds.

No, I understood just right. The launcher thing.

What I was asking about (sorry if I used the wrong words) is if FS2_Open REALLY needs the registry. Forget compatibility, this is an abstract question.

I'm just collecting data, not proposing designs for the next Space Shuttle.

As for the quote, it was the motto of a company who created another landmark game some years ago. But design and functionality go hand in hand. You can't have one without the other and you can't have a good end product without them.

[The E]

In principle, FS2_Open doesn't really need the registry. However, and this is a big however, all windows builds will need it until that dependency is discarded in favour of some sort of .ini file. As a result, IF the new Launcher is to be compatible to all FSO builds (Just as the current Launcher is), it HAS to be able to read and write to the registry.

[Goober5000]

You know, a launcher rewrite is already in progress.  It's just taking a long time.

[Spicious]

I've got a rewrite going too; it's probably going even slower. So far it's just a copy of the existing gui in wx without any functionality.

Would it really be terrible to at some point break backwards compatibility with new launcher and new fs2_open versions?

[kkmic]

Would it really be terrible to at some point break backwards compatibility with new launcher and new fs2_open versions?

Hmmm... since most of the mods released lately use a recent build, it would probably not be such a big deal. There will be some angry posts on the board though, I can guarantee you that

@Goober5000:
The idea was not to just rewrite a platform-independent launcher, but to create a new one, improving on it's current capabilities and adding new ones.

I know we are discussing starting ANOTHER launcher project, but is there anyone CLOSE to finishing one? Making our work redundant...

[Aardwolf]

I say unless we can get it set up so FRED2/FS2 will properly launch and start a mission when I double-click (or right click and select Run/Edit in FRED2/whatever), screw the registry.

That's all it'd really be good for, as far as I can tell.

That said, if something like that could be set up, I'd be all for it.

[chief1983]

I think you're misunderstanding what I said. The Launcher needs to read and write to the registry if it is supposed to be compatible to all FSO builds.
Passing parameters via text file is necessary to a) avoid storing said parameters in the registry b) Keep them around in a human-readable format and c) because of the string length limits portej05 mentioned.
Now, the only scenario where the Launcher is not allowed to write this file to a location where FSO can read it (IIRC) is on Windows Vista and later with the Launcher not being in the same directory as the FSO exe. YAL just writes to ~/.fs2_open, and I imagine Soulstorm's Launcher does the same thing.

Design is law

No. Functionality is.

Both are.  It has to work, and it has to be usable and maintainable.

[Iss Mneur]

Would it really be terrible to at some point break backwards compatibility with new launcher and new fs2_open versions?

Hmmm... since most of the mods released lately use a recent build, it would probably not be such a big deal. There will be some angry posts on the board though, I can guarantee you that

Someone correct me if I am wrong but I thought this would be most useful to support most, if not all, mods that are still available (not that it would happen right away, though) so that it very straight forward for new freespace users to get started with the game.  Yes, I realize how many completed mods there are just on hard-light, but that would be part of the designing/specing the system so that community can maintain the data that the installer offers.

Not to be presumptuous but it seems to me that Hellzed's primary motivation for this suggestion is to be able to "sell" Ubuntu and a neat shiny game, (freespace), to the francophone computer users.  That, IMHO, is something any of the linux distro projects would like.


@Goober5000 and kkmic:
I think the idea was to make the whole thing polished and user friendly, because that seems to be the only thing that does not exist the basic, polished, and shiny launcher.  Everything else already exists.  We have Turey's installer that already exists and does work, but I am under the impression that the installer as it is now does not do diffing, it just downloads the entire vp that is out of date.  In some cursory searching that according to FSOInsaller.com that Goober5000 is rewriting the installer, but the last date on that site was from 2006/2007, so the question for Goober5000 or anyone else that knows, what features does this rewritten installer have/should have and what is its status? I, like quite a few coders, am quite lazy and if someone has done some of the work already, I am not going to re-implement it just because I can.

I know we are discussing starting ANOTHER launcher project, but is there anyone CLOSE to finishing one? Making our work redundant...

I too would like to know.  I see there are four launchers that have been released, (again cursory searching), 1 for win32, 2 for linux, and 1 for MacOS X.  Seems YAL does both windows and linux, not OSX.

Modifying an existing launcher is not out the question, but this is more of a design type document for a "basic, polished, and shiny" launcher that will integrate a an installer so that it can check for and install updates to the mods, engine, maybe fred and itself.

[Goober5000]

In some cursory searching that according to FSOInsaller.com that Goober5000 is rewriting the installer, but the last date on that site was from 2006/2007, so the question for Goober5000 or anyone else that knows, what features does this rewritten installer have/should have and what is its status? I, like quite a few coders, am quite lazy and if someone has done some of the work already, I am not going to re-implement it just because I can.

The 2006/2007 date is when Turey stopped updating the installer.  Recently he delegated the code to the SCP, and I started a general code cleanup about six months ago.  It's reasonably close to being done; I just need time to finish it.

I know we are discussing starting ANOTHER launcher project, but is there anyone CLOSE to finishing one? Making our work redundant...

I too would like to know.  I see there are four launchers that have been released, (again cursory searching), 1 for win32, 2 for linux, and 1 for MacOS X.  Seems YAL does both windows and linux, not OSX.

Indeed.  I figured it would be best to start with the most fully-featured launcher and make it cross-platform.  This I judged to be the Win32 launcher that was originally created by RandomTiger and is now distributed with FSO.  I am also performing a code cleanup on the launcher in parallel with Turey's installer.  Once I finish, the plan is to hand it off to portej05 who will oversee conversion of the GUI to wxWidgets.

[Iss Mneur]

The 2006/2007 date is when Turey stopped updating the installer.  Recently he delegated the code to the SCP, and I started a general code cleanup about six months ago.  It's reasonably close to being done; I just need time to finish it.

Hmm, interesting. Out of curiosity (no I have not checked to see if I can find the code) are the features that we are discussing here planned/implemented? In particular the diffing of .vp files, the ability to check for updates quickly, and to apply updates only to a specific mod. Is the installer also being ported to C/C++ and wxWidgets or is it staying in java?

I know we are discussing starting ANOTHER launcher project, but is there anyone CLOSE to finishing one? Making our work redundant...

I too would like to know.  I see there are four launchers that have been released, (again cursory searching), 1 for win32, 2 for linux, and 1 for MacOS X.  Seems YAL does both windows and linux, not OSX.

Indeed.  I figured it would be best to start with the most fully-featured launcher and make it cross-platform.  This I judged to be the Win32 launcher that was originally created by RandomTiger and is now distributed with FSO.  I am also performing a code cleanup on the launcher in parallel with Turey's installer.  Once I finish, the plan is to hand it off to portej05 who will oversee conversion of the GUI to wxWidgets.

Interesting. Is the clean up a cross platform conversion? or is more a remove the unclear code and do the cross-platform stuff when converting to wxWidgets?  The reason I ask is because wxWidgets, as you probably know, has a large amount of cross platform tools in addition to the GUI stuff itself, in particular file system stuff.

Anyway, the original idea of this thread was to have a very simple and pretty launcher to present to new users and so far this seems to be something that we have will have to do.  But the basic launcher would need an advanced settings capability that could be the existing fully featured launcher, which IMHO, would work out quite well.  Once we figure out what we want to do for the new shiny launcher, we will come knocking.

[Goober5000]

Hmm, interesting. Out of curiosity (no I have not checked to see if I can find the code) are the features that we are discussing here planned/implemented? In particular the diffing of .vp files, the ability to check for updates quickly, and to apply updates only to a specific mod. Is the installer also being ported to C/C++ and wxWidgets or is it staying in java?

The first two of those aren't planned.  And no, I'm not rewriting it in C; I'm keeping the Java code but cleaning it up a bit.

I'm also not working on it full time; I'm sort of interleaving it with the launcher and all the other projects I have, which is why it's taken a few months.

Interesting. Is the clean up a cross platform conversion? or is more a remove the unclear code and do the cross-platform stuff when converting to wxWidgets?  The reason I ask is because wxWidgets, as you probably know, has a large amount of cross platform tools in addition to the GUI stuff itself, in particular file system stuff.

It's a multi-stage process.  Cleanup comes first, then cross-platform compatibility.

[Soulstorm]

Well, lately, I had the urge of rewritting the FS2_Launcher for OS X. Since its release, my programming skills improved, but not the code base, and I reached a point to say "what the heck was I thinking when writing this part?". So, when I began rewriting it, I created this thread to ask for feedback and features request.

So, people pointed me towards creating a cross-platform Launcher. Seems the need is so great for a cross-platform Launcher that I can hardly find any reason to start a new project. For the Launcher on OS X even if I had some cool ideas, like implementing some eye-candy, updating some 3rd party mods, mini-browser for the site for each mod, connection with the Freespace Wiki for explanation for the command line arguments, and stuff like that.

I understand that although a native program for any platform can go deeper into that particular platform's features and use them, and it will have a better look-and-feel with less effort, I also know that a community needs ONE codebase, since everyone must be safe that if someone leaves the project, the rest of the community will have the hardware-software necessary to continue his/her work.

Therefore, I will conform with the rest of the community. I would not like to be the stubborn one that refuses to go with the flow and create platform specific Launchers

So, if anyone wants my help for any cross-platform Launcher project, my C++ skills are available

[karajorma]

Excellent Soulstorm.

So, people pointed me towards creating a cross-platform Launcher. Seems the need is so great for a cross-platform Launcher that I can hardly find any reason to start a new project. For the Launcher on OS X even if I had some cool ideas, like implementing some eye-candy, updating some 3rd party mods, mini-browser for the site for each mod, connection with the Freespace Wiki for explanation for the command line arguments, and stuff like that.

So we just have to make sure the cross-platform one can do all that too.

[chief1983]

Yeah there's not much preventing that.

[Iss Mneur]

@Soulstrom: That's great.

@kkmic or @Hellzed: Has there been any progress on this?

It it now October and I now have some free time.

[kkmic]

I've sent Hellzed a PM. We should wait for his answer.

Also, I'm curious what the SCP Meeting will produce Perhaps we should  wait for it to take place and conclusions to be drawn. Though his means another month kinda lost...