[mgt-devel] [Multignometerm-devel] work plan

Mohammed Sameer msameer at foolab.org
Sun Feb 17 19:37:11 UTC 2008


On Sun, Feb 17, 2008 at 12:50:27PM +0100, Emil Nowak wrote:
> Hello,
> What is plan of our work?
> On BTS you proposed two possible solutions:
> 1) Write MGT from scratch (porting features)
> 2) Migrate current MGT code (gtk1, libgnome) to gtk2
> 
> Please tell me more about pros and cons for each solution that you discovered
> so far.
> 
> I saw your tabterm project it looks quite good. However if You are using C++
> it will be better to use use gtkmm instead of gtk directly - it will make the
> code cleaner.

Hi,

I'm honestly not sure which direction to follow.
Rewriting from scratch means we will have to carefully list all the features
implemented already be mgt (I don't know all of them honestly) and re-implement them.

mgt code is really bad. Check it. The author re-implements GtkNotebook for example.
He copied libzvt and modified the code. We also have the problem of "porting" the
glade files to glade2 (Gtk2).
Problem with rewriting it is we will have to go through a lot of coding and debugging.
It might be the same effort as porting it but I don't really know.

How's your gtk1/gtk2 experience ?

Another problem is that libzvt2 (mgt used zvt1.x as its terminal widget) seems dead.
We can of course adapt and maintain it. It's scheduled for removal from Debian and
it'd be easier for us to maintain it rather than re-adding it.

VTE is the new widget used by GT. I noticed it's a bit slower. That's why I'm not sure
which direction to follow.

As for tabterm, I cooked it in a very short time for it to be my MGT replacement.
Of course we can use it as a codebase for MGT2. Problem is depending on gtkmm (I
personally love gtkmm) means we need to pull a lot of dependencies. and if we decided
to use zvt, we will have to wrap it ourselves.

I like the C++ typesafety, I like sigc typesafety too. That's why I used sigc to
"communicate" between classes. The "glue" between gtk signals and sigc signals is a bit
ugly I have to admit but it can be implemented using a C++ template.

Which preference do you have ? C or C++ BTW ?

What do you think about all the above brain dump ?

Cheers,

-- 
GPG-Key: 0xA3FD0DF7 - 9F73 032E EAC9 F7AD 951F  280E CB66 8E29 A3FD 0DF7
Debian User and Developer.
Homepage: www.foolab.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/multignometerm-devel/attachments/20080217/34e764f1/attachment.pgp 


More information about the multignometerm-devel mailing list