@EntilZha
From the porting section of the OOo website:
Quote:

Platform-Dependent Implementation The most difficult part about platform-dependent implementation is language binding, since it requires some knowledge of the CPU architecture and how the compiler constructs vtables and so on. This is a time-consuming activity. For existing ports, this is implemented in the module bridges. Loading Shared Libraries Another area where some development might be necessary is the loading of shared libraries. The source code assumes the availability of dlopen() and associated files. For example HPUX 10.x uses something like shl_load() and names shared libraries differently, however the necessary changes should be minor.

I don't really know anything about the first one. We will be using GCC to compile C/C++ on PPC. As there is a port of OOo to LinuxPPC port allready which will use GCC how much work will be requiered here if any.
With number two dlopen is part of the IEEE Std 1003.1-2001 (POSIX v3) standards. How well does OS4 support this? Is it likely to meen a lot of tweaking?

@EntilZha

Don't know about text (Well ATM I don't know much about OOo internals ). I know OOo can do it but I wonder if they have their own text functions or want to use OS ones?

Last edited by digitaldisaster on 06-Jan-2005 at 12:51 AM.

I voted no, because although I would like to muck in I don't have the time and don't particularly want to sit down at a computer for long after working 8 hours a day programming. Plus my knowledge of programming for the amiga is ass.

@digitaldisaster
Quote:

What sort of overhead does AGG impose?

You really should check it out in more detail. AGG is more than just a 2D gfx library. It is a framework on which you can build a gfx subsystem. Some people have already added h/w acceleration as well. It is ideal for this kind of project IMHO but I'm not sure if it has all the necessary features but neither does graphics.library...

_________________
ExecSG Team Lead

@digitaldisaster RE: Dependency graph

Yep, it's evil. Probably very similar to Mozilla. It's pretty much known to be a beast. The Java issue mentioned appears to only apply to the installer bits, although I _thought_ there were other Java bits in OO.

On the plus side....if the OS abstraction layer is ported, that by itself could become a very useful tool in porting other software, much like the Mozilla and Apache portable runtime/cross-platform libraries are re-used elsewhere....

_________________
Are we not done with the same silly arguments and flames yet??!

@Colin_Camper

Quote:

How come that SkyOS guy has got all these ports going? He's got Firefox, OpenOffice, The Gimp etc etc - The SkyOS community must be way smaller than the current Amiga, surely.

Yes, but SkyOS, unless I'm sorely mistaken, also implements a POSIX API, which makes life infinitely easier to port at the 'system level' (process creation, file manipulation, IPC, pipes, etc, and then if POSIX threads/pthreads are also implemented, life gets easier yet.

For most apps, they build on top of 'something'...with POSIX underpinnings, it's MUCH easier. That's also, BTW, how Linux managed to get so much ported to it initially, as Solaris, HP-UX etc are/were 'mostly POSIX compliant' at the time, so the actual porting work was perhaps 25% of the code versus 80% or more.

_________________
Are we not done with the same silly arguments and flames yet??!

@olegil

Quote:

What I would REALLY like to see is QT + Opera. Think about it. If a group of programmers could approach Opera Software ASA with something like "Hey, we need your browser. We have already ported OpenOffice.org, and we're not afraid of NDAs".

While I like Opera personally, and used it cross-platform as my main browser for a while, I always tend to move back to Mozilla/Firebird. The biggest gripe I have with Opera, actually ANY company that does this....is they charge per platform, without offering a multi-platform bundle/discount. I would have no problems paying $ for something I truly need, but few companies 'get it,' even when they're dev tools, so should in theory know better. I wanted a Python capable IDE recently (aside from emacs/vi), primarily for a decent debugger, and ran into the same issues there- either it only supports one platform (umm, which is silly for langs like Python and Java), or you have to pay per platform. I want the same tools/apps on every major platform I need to use that functionality on....and would pay for it for some apps, but not at $100 per platform flat rate, even if only for Linux, Solaris, and Windoze (what I needed/wanted an IDE for, and Opera)

QT has licensing issues with free software, which actually affected one of the few IDEs I found...I think it was Komodo, but may be mis-remembering which one it was that used QT. The Linux version was free (as is QT), the Windows version required a commercial license which isn't overly inexpensive. The other negatives there was it didn't handle syntax other than python, nor support Solaris....which is why I stay with Eclipse.

As much as I like _some_ apps on KDE (which uses QT as it's underlying widget lib), and am not overly fond of GNOME....GTK would be a more sensible port (and I _think_ opera uses currently, but staticly linked?)

Quote:

Other programs are also possible once we get onto the right track. It's all about starting in one end and keeping it up until we are satisfied

Yeah, I think you're dead on on this....with a decent 'portability layer' (which OOs may be) and a few 'major' apps (email, word processing, IM client, Media players), an OS can become 'useable enough' for the 'everyday person'....at which point AOS/Amiga has some sort of chance of expanding (we won't talk about taking on other OSes, just expanding from where it is now)....and if those apps are 'identical' to the ones people are now used to using, it makes it that much easier. Get the apps, add marketing and some good publicity, and perhaps a few good 'big' deals, and there's a start- have Eyetech make a deal with Kinkos, airports, Starbucks, or somewhere similar for 1000 uA1-Cs packaged in a small case for $600 each, and use them as a 'kiosk style system'....

_________________
Are we not done with the same silly arguments and flames yet??!

@EntilZha

Quote:

Python should not be a problem... Hans-Jörg grabbed the sources and compiled the core of it... A bit more work is needed for the rest of the stuff, but hopefully nothing too serious...

What version of Python?

_________________
Are we not done with the same silly arguments and flames yet??!

@wegster

Yep, as I said earlier a combenation of the strong POSIX underpinnings and the fact that the main developer, Robert S, has intermite knowledge of the internals and access to the source code to change them as and when he pleases

@wegster

There are Java bits in OOo to enhance its functionality but they are optional extras and can users can surive without them. The critical Java problem is that it is used to manipulate xml files during the build

@wegster

Even if you can't reuse the code a complete project on the scale and of the quality of OpenOffice.org shows that you have the experience and the knowledge nessacery to port closed source commercial software such as Opera

@wegster

Quote:

The biggest gripe I have with Opera, actually ANY company that does this....is they charge per platform, without offering a multi-platform bundle/discount.

Actually, they recently changed to a 1-license-any-desktop policy. Right now if you
download a new version (http://www.opera.com/download/) you can buy a desktop license for $24, which includes a free upgrade to the upcoming 8.x version.

Scott Woodall
(who is not getting kickbacks or anything)

@EntilZha

I am just an end user, but I would gladly contribute some money toward such a project.
As much as 1K US.

Tom

So what's next...I voted yes, but no time, but if I read some posts here, this project might have a real chance.

So, who's setting up the servers ? What's next? Let's start it off :)

I would be very interessted in participating but I think it wouldn't be fair to be a part of it...I start to many things just to find out I have no time...but if something is get going, I'd like to take a look now an then, and maybe there are small enough portions I can contribute to.

_________________
AmigaOne XE G3/933/VIA/FM801/R200 (fixed),
G3SE/600/Voodoo3/Sil680/RTL8139/SBLive! (noiseless!)
µA1-MK2/G3/933/R200/CMI8738
XE/G4 (broken 7450/800)

@swoodall
RE: Opera multiple platform

Cool, that is new, or to me at least. Pricing on site says $40 for multiple platform, but still good either way, and it include OS X _and_ FreeBSD...very cool. I'm hoping they 'just get it' instead of this meaning they aren't doing well financially though...

@thread RE: site space-
Were someone to put together a 'semi-official team' and a project plan of sorts, it's highly possible to get space on the OO site.

@thread RE: Project Management
I've done this before, in 'real life' as well as in an online community with 20-40 active developers). It's not overwhelmingly glorious work, always takes more time than you expect it to (heh, sort of _exactly_ like most coding projects! ), can be tedious, and to be _good_ at it, expecially in a more 'informal' setting, it really does help to have some technical skills, and to naturally be, and enjoy, paying attention to lots of small details that no one else does. It also means you've got to really see 'the big picture' while individual devs focus in their 'small' parts of 'their world' on their specific code pieces. It also means that if you like to try to 'push' people as a way of 'management,' you may want to think again, as it doesn't work well in the traditional sense with a group of unpaid volunteers (that IS actually meant as a serious statement than some 'management' just may never get)

Doing PM on a project like an OO port _will_ be a pretty big chunk of time, especially initially. Think of this as ranging from a part time 20 hours/week to a full time PLUS job. Each group has it's own dynamics, and some devs can, and occasionally do, produce good docs and plans for their part of a project. Most don't, but can be coerced into giving a verbal description over the course of IRC or several emails or *gasp* a phone call. So, the group itself really determines just how much work falls into a PMs lap versus being shared.

What I'd expect to see is a small group of people taking 'initial ownership' of the project...say, an admin, a 'docs' and/or website guy (or both), and a few developers (say 3-4). Hash out 'the basics'- who hosts site...if not OO then who, what domain, forum-yes/no, source control and defect tracking system used, build environment and tools/versions, source analyzer, etc...then go through existing OO docs on porting, APIs, individual layers just to point of seeing just how much 'isn't known yet and needs to be.' A few logged IRC meetings to get things started, then 'official' tasks to be defined to get the basics of a master tasks list with 'major tasks' listed along with up to a paragraph of notes by each person doing the initial investigation. Most of these should be links to either the OO site, or to new docs (or wiki entries) on the 'Amiga OO site'. NO CODING YET! Get everyone to agree at this point to the state everything is in and agree the major tasks (not 'deep' implementation details, although references to code/places/sites/tools that could help or that may work out is good). site admin gets a backup, and ideally no one else has root/admin privs on the site box (this becomes important when inevitably someone gets pissed off...)

Now...if someone in the original group actually thinks they _want_ to go to 'the next level' (up, down, sideways? ) and is capable of doing PM work, possibly to the exclusion of his/her initial tasks....let them, otherwise bring in a PM, preferably one that can help out a bit on technical bits (including docs, strategy etc). Bear in mind if it's anticipated that a PM isn't going to be from the initial 'technical group,' whatever doc storage, content management system, wiki, etc and web tools are available should be chosen based on someone being less technical...they _will_ need to publish docs, easily, and others will need to be able to refer to them (which makes a CMS, wiki or similar a not so bad idea...)

The other option is to bring in a PM prior to the initial port investigation and to outline 'here are the options' for tools, site stuff, doc formats, etc...and let him/her make 'official' contact to the existing OO team to see about web space, take a look at the tools that will affect him/her etc and then go from there. The primary reason I'm opposed to having an 'official PM only' from the start of a project like this is that while a lot of stuff is in the details- subversion or CVS, or Perforce? do all integrate into whatever IDEs, if any, in use by most devs? which CMS? etc...some PMs feel the need to 'have their hands in' when they're more technical decisions than an 'I want this shopping list.' Nobody (well ok, few people) want to spend a huge amount of time in getting the basic infrastructure up by having to write code to add new features to perhaps a poorly chosen CMS, forum etc...

Anyways....it's a fair amount of work by all involved...and many projects have failed to even get to this point, or abandoned after making some initial progress. (another reason to make sure all work is shared via site/source control/shared tools, and for sysadmins/web guys to have backups, as well as a 'real' backup of work done..)

We now return you to the regularly scheduled single sentence comments

_________________
Are we not done with the same silly arguments and flames yet??!

@EntilZha

Isn't there a port for 68k of Python already?
Could it be used, or at least the author informed, IF it gets reality?
He might help, he knows the code already?

@EntilZha

Yea, they can be pulled from CVS, i only downloaded these out of curiousity

_________________
Amon's digital home

@EntilZha

Long time no message from me. I had to vote with "yes, but have no time". However, I'll work my a... off to get JAmiga finished, so that you can eventually use if for both, building and the extra features it brings.

Regards,

Peter

@digitaldisaster

I said we should start by porting SVN for a reason, you know

Once you have the source and just want to check if there's anything new, you'll see why I don't like CVS...

_________________
This weeks pet peeve:
Using "voltage" instead of "potential", which leads to inventing new words like "amperage" instead of "current" (I, measured in A) or possible "charge" (amperehours, Ah or Coulomb, C). Sometimes I don't even know what people mean.

@digitaldisaster

I had writen a set of functions that mimic dlopen/dlclose/dlsyn, at least to a certain degree. Since OS 4 binaries are ELF file and can contain symbol information, this shouldn't be too much of a problem.

We also recently added baserel-addressing to the compiler, so that a llibrary can be compiled with a per-user data segment if this is required.

So the infrastructure is there, it just needs a bit of additional work.

_________________
Seriously, if you want to contact me do not bother sending me a PM here. Write me a mail