From: David Doolin (doolin@cs.utk.edu)
Date: Thu Jul 06 2000 - 13:24:36 EDT
[snip snap]
>
>I am concerned that once you start becoming dependent on msvc,
>you get into situations like what has happened with ActiveState
>Perl. Now, I have nothing against ActiveState. They've done
>an _excellent_ job. I am concerned that their toolset having
>unique MS-Windows utilities and options could end up eventually
>forking the Perl code base and world. MS gives ActiveState
>funding and MS has tried to fragment the Java through embrace
>and extend.
Absolutely. Don't want critical dependence in on *any* os as far as
I am concerned. So far, Daniel has done an almost unreasonably good
job. For instance, relying on lex/yacc/flex/bison to compile would
be a complete show-stopper for me, and probably a lot of other win32
people. Not that lex/yacc is bad, they are great in fact. They just
don't ship with msvc (unfortunately). So there is some extra effort
in the parser on that count.
>
>I hope we can agree that any changes necessary to support Win32
>would be completely compatible from the API standpoint with the Unix
>code base.
Absolutely. That's why I am using the raw win32 api instead of
mfc in my current code (it's not that bad btw, considering its
age and the amount of backwards compatibility supported).
Having a completely api compatible across win32 and unix is
in fact the point. Anyone wanting to use MS specific xml
features really should just link to the MS xml parser anyway.
>
>If the Cygwin environment works as intended, there wouldn't be
>any changes from the Unix sources necessary. Of course, this
>is not always possible and a lot of sources "ported" to work
>under Cygwin have changes that conflict with the msvc
>compilers, often involving incorrect use of "#ifdef WIN32" when
>a Cygwin specific #ifdef was what was required.
>
>I completely agree that if libxml were made to work with
>Cygwin that care should be taken not to conflict with msvc
>changes.
I like the idea of libxml compiling under everything.
>
>I'm also in sympathy with Erwin Rol's concerns surrounding
>Cygwin. In many ways it is an immature environment and I
>wonder if it will ever be useful for programs upon which
>people can depend. Although, you can distribute binaries
In my case, I am collaborating with several others civil engineers,
some of whom are using fortran on a parallel code base. It really
is unreasonable to expect them to master the intricacies of
the unix build system. Much better for them to concentrate
on writing correct code and doing correct engineering. The purity
of the engineering overides the purity of the build
environment. Having written makefile and done a bit of autoconfiscating
myself, it just is much easier to use the appropriate tools
for the platform. On windows, this will be an ide for
most people.
>built on Cygwin with a DLL that doesn't require one
>to install the compiler and environment. Also, there is
>the MINGW32 set of tools, which is the GNU tools and
>compilers using the MS C DLLs. This is another flavor
>that might be investigated.
>
On the other hand, I really like gcc, and frankly, I trust it
to be a better iso c compiler than msvc. Which is why I
periodically spend a few days blasting my numerical kernel
through gcc. The xml being essentially platform independent
is a big boon here.
>>
>> Also, nothing personal here, so don't take it so...
>> For those of us limited to ms tools, not having libxml
>> means linking ms xml parser.
>> So there is a bit of give and take here.
>
>This is a _very_ good point.
libxml can be a "compelling" app for windows developers.
The more Windows developers that use libxml, the more
Windows developers will use xml. It will bootstrap itself.
The main contribution I think I have made is not at all
in the libxml code base, nor in its use, but providing
the msvc ide files. My goal has been someone should be
able to dl what I posted, double click the workspace,
build it, and use it. Eventually, I would like to see a
win32 subdirectory in the cvs sources for any ide:
msvc, borland, whatever. Make it really *easy* for
a win32 developer to use it, so easy that they have no
excuse not too. FWIW, its pretty much sop in the Windows world
to distribute ide files with ssource (and there is a
considerable amount of free example code running around the win32
world, I would be hosed without it).
>
>I think we're in agreement here, but I would like to
>point out that having options surrounding using libxml on
>Win32 environments (msvc built, Cygwin, Borland, etc.) is
>also a good in that it gives people who wish to
>use libxml in Windows environments more flexibility.
I agree; we are thinking more or less along the same lines.
-dmd
---- Message from the list xml@xmlsoft.org Archived at : http://xmlsoft.org/messages/ to unsubscribe: echo "unsubscribe xml" | mail majordomo@xmlsoft.org
This archive was generated by hypermail 2b29 : Wed Aug 02 2000 - 12:30:22 EDT