Note: This archive is from the project's previous web site, ethereal.com. This list is no longer active.
Yes, and in the spirit of the GPL, I'd rather see the dissector source code checked into Ethereal's CVS. This runtime linking, while technically possible, would create an avenue for modified versions of Ethereal which this development group does not have all the source code for. Imagine the support questions this list would get. The first couple of email exchanges would be needed to sort out which modules the person is using.
What's wrong with compiling and linking the whole product? It's not really that hard. If someone can't do that, I doubt they could compile a module correctly.
Is waiting for the next release the hard part? Perhaps this group should adopt the Linux release scheme: Even numbered minor versions are stable releases, and occur as often as normal releases do now. Odd minor numbered releases occur every few days, and are experimental. Users are encouraged to use the even numbered ones. Developers and early adopters (super users) are free to try out the experimental releases.
My suggestion is to release version 1.0, and version 1.1 at the same time. Then build the 1.1.X branch often, and only release the 1.0.X branch when it's been stabilized and tested. Cross-port only the successful changes from the 1.1 branch to the 1.0 branch, after they have been well cooked by the developers. This way, people who need fast turnaround on new dissectors can get it, while not disturbing the stable release pattern that end users depend on.
-----Original Message-----
From: Chris Waters [mailto:chris.waters@xxxxxxxxxxxxxxxxxxxx]
Sent: Wednesday, December 18, 2002 8:49 PM
To: EtherealDev
Subject: Re: [Ethereal-dev] Easier dissector building
Hi,
You still need to compile the shared library/DLL. Under Windows this
requires you to have MS VC++. Not many network administrators are likely to
have this.
Regards,
Chris.
----- Original Message -----
From: "jharris" <jharris@xxxxxxxxxxxxxxxx>
To: "Chris Waters" <chris.waters@xxxxxxxxxxxxxxxxxxxx>
Cc: <ethereal-dev@xxxxxxxxxxxx>
Sent: Wednesday, December 18, 2002 6:40 PM
Subject: Re: [Ethereal-dev] Easier dissector building
> You dont need to recompile ethereal. If you compiled the new C code as a
> shared object and then load and initalize it at runtime, you could leave
> ethereal alone. As the disectors already dynamically register themselfs
> at run time this would be rather easy. The trick as with all dynamic
> loadables is keeping the versions straight.. and the apis consistent.
>
>
> On Wed, 18 Dec 2002, Chris Waters wrote:
>
> > Hi,
> >
> > Rather than building a translator like idl2eth which will create C code,
how
> > about a runtime/end-user method of adding dissectors. The problem with
any
> > scheme that generates C code is that you still need to compile Ethereal
to
> > use your new dissector. Unfortunately compiling Ethereal is not trivial
on
> > some platforms (e.g. Windows) and anyone who can get it to compile
probably
> > could have written their dissector directly in C anyway.
> >
> > Both Analyzer and Airopeek have a meta-language which is used to
describe
> > protocols and users can add their own protocols simply by providing
files in
> > this format. I have some colleagues who would have no idea how to
compile
> > Analyzer, but they have managed to add dissectors for the Bluetooth
protocol
> > suite using the built in protocol-description-language.
> >
> > The drawback of a non-compiled solution is speed, but personally I would
> > give up some speed for the advantage of being able to quickly add a
> > dissector for a new protocol.
> >
> > My preference would be to use a scripting language, such as lua
> > (www.lua.org), to describe the protocols. It is well suited to
describing
> > data structures, such as simple arrays of fields, but because it is a
> > scripting language you can write as much or as little code as you like
to
> > handle complicated protocols.
> >
> > Regards,
> >
> > Chris.
> >
> > ----- Original Message -----
> > From: <andreas.sikkema@xxxxxxxxxxx>
> > To: "Jason House" <jhouse@xxxxxxxxx>
> > Cc: <ethereal-dev@xxxxxxxxxxxx>; "Ph. Marek" <philipp.marek@xxxxxxxxxx>
> > Sent: Wednesday, December 18, 2002 9:27 AM
> > Subject: Re: [Ethereal-dev] Easier dissector building
> >
> >
> > > > It's definitely do-able to have an automated tool to write
dissectors
> > > > for you.
> > >
> > > Exactly. I have adapted an ASN.1 compiler to generate code that
> > > can be used for an Ethereal dissector. It took me a couple of
> > > months of hard work, but you can do it ;-)
> > >
> > > And then it still needs some manual work....
> > >
> > > --
> > > Andreas
> > >
> > >
> > > _______________________________________________
> > > Ethereal-dev mailing list
> > > Ethereal-dev@xxxxxxxxxxxx
> > > http://www.ethereal.com/mailman/listinfo/ethereal-dev
> > >
> >
> > _______________________________________________
> > Ethereal-dev mailing list
> > Ethereal-dev@xxxxxxxxxxxx
> > http://www.ethereal.com/mailman/listinfo/ethereal-dev
> >
>
> -- James Harris
>
>
_______________________________________________
Ethereal-dev mailing list
Ethereal-dev@xxxxxxxxxxxx
http://www.ethereal.com/mailman/listinfo/ethereal-dev
- Prev by Date: [Ethereal-dev] Re: Re: compile problem by using VC++
- Next by Date: Re: [Ethereal-dev] hanging graphs
- Previous by thread: Re: [Ethereal-dev] Easier dissector building
- Next by thread: [Ethereal-dev] compile problem by using VC++
- Index(es):