[khtml-win32] Re: Offer to Help with Konqueror port

[Alan G. <aj...@iz...>]

* Charles Paglee <ch...@ca...> [2003-03-10 06:40]:

> On Fri, 7 Mar 2003 23:13:50 -0600, "Alan Gutierrez" <aj...@iz...>
> said:
> > * Charles Paglee <ch...@ca...> [2003-03-07 22:52]:
> > 
> > > A group of engineers at one of our partners in China is interested
> > > in working with you to finish the port of Konqueror to Windows.
> > > Would you like some help in this project or are you almost
> > > finished. It seems like you were moving at a pretty rapid pace and
> > > then stopped all of a sudden.
> > 
> > > If you could just drop me an email to let me know you received
> > > this email I would appreciate it.
> > 
> > I'd greatly appreciate any help. Things were moving rather quickly,
> > but my day job has absorbed all my available computer time through
> > Feburary. Just recently I started to pick the code up again.
> > 
> > It would help to have anyone else look at the code as I work on it.
> > It is quite a port. Much of the Apple code is a wrapper around
> > NextStep and Objective C++, so there is a lot that has to be
> > written. It requires solutions for networking, Unicode, mime, fonts,
> > and windowing. It would be nice to delegate some of these. If they
> > are Chinese engineers, they are liable to have far more Unicode
> > experience than I do, for example.
> > 
> > Not that there is really that much left to do, I was surprised at
> > how few files were left to implement.
> > 
> > Please introduce me to your developers. I'll work on getting the
> > source code into CVS.

> I work extensively in China and have known this team that has needed to
> include HTML functionality with their product for some time. Unicode is
> definitely something they can tackle pretty painlessly. I'll be going
> back over there in a few weeks if that is quick enough for you.
> 
> When do you expect a 'Hello World' alpha to be ready?

Mr Paglee

I'm only now starting to resume my development of the KHTML Win32
    port. I don't know what to say. I'm just beginning to think
    about this again.

I am going to rush towards Hello, World this time, rather than port
    a class at a time. The sooner people can donwload and compile
    *anything*, the sooner I'll start to garner the interest of
    other developers.

Since you're listening, let me say that part of the reason that this
    project fell by the wayside was because I am a UNIX developer by
    day. I'm been heavily involved in UNIX devel for the past two
    months. The progress I was making was arrested by my
    unfamiliarity with the VC++ IDE, and a lack of desire to learn a
    completely different method of project management (especially
    when I've inveseted so much effort in learning autotools last
    year).

What I'd like to do at this point is generate dozens of little test
    programs, for each little component that port successfully. This
    is what I do in UNIX. Create a main that returns 0 if the test
    succeeds, then glue the tests together with a shell program.

Having dozens of little build targets in VC++ is a nightmare. I
    suppose a VC++ developer would create some sort of monolithic
    test program, or a com interface. I don't feel like managing
    that complexity for a test suite.

Also, the compiler I've got, VC 6, has terrible C++ support. It
    generates copious compiler warnings that simply have to be
    disabled.  It's not reassuring to have a compiler that can't
    understand the template names it generates. I've been doing much
    C++ hacking in the UNIX env lately using gcc. The ability to use
    standard library template classes would eliminate a *lot* of the
    KWQ code.

Here's the direction I'm going to take:

1) Install cygwin and use make, autoconf, automake, bash, perl as a
    build envrionment.

    Down the road, when KWQ compiles into one big chunk, it ought
    not be difficult to create a VC 7 project that will also build
    the application. This is the sort of thing a Win32 volunteer
    could contribute.
    
    Yes, it would be nice if Joe Win32 developer could just download
    khtml-win32, unzip and start hacking. Still, I've not received
    any offers of help from Win32 deveopers, so I may as well
    accept the help offered by GNU and Cygwin. It is going to be me
    alone for a while, so it is best I don't add considerably to my
    study tasks.
    
    Also, starting out, it is hard to say where help will come from.
    There might be others out there who straddle OSes who want to
    hack with gcc, or bcc. GNU will allow people to use the compiler
    they have.

2) Investigate using Comeau C++ as the development compiler.

    Comeau C++ generates C that is fed to the Visual C compiler. It
    will still allow visual debugging of the code, but it is command
    line oriented, and it supports put near all of ISO C++.
    
    I feel that I can get a lot more done if I do not have to add
    compiler debugging to my programming tasks. Also, the KHTML
    hacks will merge nicely with the existing KHTML code base. There
    will be fewer #ifdefs to support Win32.

    VC++ 6 really is a broken compiler.

    Comeau C++ is $50 and it can target all the common Windows C
    compilers VC6/7, BCC, GCC, more.

    I might impose this $50.00  barrier to entry to the project, but
    that is a lot less than the $n00.00+ barrier set by VC7.

    Finally, using Comeau will make khtml-win32 clear of any
    anti-GNU oddities in the MS EULA.

Thank you for your support and persistant nagging.

-- 
Alan Gutierrez - aj...@iz...
http://khtml-win32.sourceforge.net/ - KHTML on Windows