Re: [gDesklets] FYI discussion of a merger

[Bjoern Koch <H Humpel gmx de>]

Hi there!

To be honest, I am *very* skeptical about this.
True, there might be a few nice advantages which can/could be made, but
mainly this doesn't quite aim for the main "dilemma" gDesklets is in
(and has been for the last few month/years).

WARNING! I am going to overdue a few times in the following text to
point things out. I don't want to be annoying, offending or whatever!
Everything is just my point of view.

> sounds like a nice idea, although someone argued that gdesklets has 
> its own (XML) dialect which results/resulted in having only a limited
>  amount of desklets (as far as I understood).

Yeah. I can't agree to this argument either. ADL is a beauty (IMHO).
If you want to write your own screenlet you will need to know some
basics about python, and if you do so, ADL won't be *any* problem.
I'd even say that with ADL you might even write a nice Desklet
without knowing too much python!

> According to http://gdesklets.zencomputer.ca/ there are about 180 
> desklets out there which -- in my humble opinion -- is quite a 
> number.

So true! 105 of the 179 desklets on the archive site have been marked
(more or less) working. 105 !
The problem is not that there aren't enough desklets out there but that
there are only 51 of them on the official gDesklets site (including 
Controls).
Even worse: gDesklets' "best desklet" - by the way: written by the
gDesklets team! - hasn't even found its way to the new website (which
has been online for 14 month now!): SideCandy.
There are 23 (!) additional SidyCandy desklets out there, and only ONE
of it has found its way to the website!
What would a user think about it ?? Looks like we have no trust in our
own desklets!

> screenlets on the other hand is just python + gtk + cairo (AFAIK). 
> The question remains to develop a common -- let me put it the 
> gdesklets way -- ADL.

yes... a common - let us name it - ADL *is* a beauty, isn't it ?

> I'm all for it, i.e. having a shared/maybe common codebase. 
> Especially, since there're a lot of ideas for 0.40 and the prototype 
> shows some basic features of the next major release.

gDesklets main problems are more to find in the way a "user" is going to
see the whole gDesklets project (and upcoming developers bascially start
as users). So the following will be from a (new) user's point of view:

1. News/Updates
16 news entries in 14 month. And most of them are about the website and
IRC meetings. If you take a look at the news from a user's point of view
it basically comes down to 3 or 4 "interessting" news in 14 month.
Same for the desklets. Not really much going on here?!
If you would be a new user (even to linux?!) and come across this site
and see there is basically nothing going on... what would you do if you
were looking for desklets/widgets and have other alternatives to choose 
from ?

2. User "interoperation"
So, you finally decided to give gDesklets a try. If you are lucky (or
should I say unlucky) and have this totally outdated gdesklets-data
archive installed with 90% of the desklets not working ... yeah...that's
fun.
So, maybe you want to ask someone what to do next to get things
working... aehm... hmmm...
IRC: noone hardly every there...
forum... forum ?? Comeon! EVERYONE uses a forum for little questions 
like this!
Aha... a Mailing list... what the f*** is a mailing list ?? OK, look up
on google... oh dear! I need to subscribe to ask a simple question and
then I am going to have lots of new eMails in my Inbox about stuff that
I am not interessted in ?? You gotta be kidding me!
Aha... launchpad... ouff... that looks a lot like "for developers only"
to me. I don't have a bug... nor a clue ;).
And: you gotta be a very persistent user to get as far as I have
outlined! Most new users would have de-installed gDesklets after they
found out that the data-package is not working or they couldn't find a
forum.
And if you finally figure out that the gdesklets-data is outdated and
even the gDesklets team tells you NOT to use it... well... everything
looks soooo dead!

Here starts the "evil circuit", because the upcoming developer - who 
comes across gDesklets as a user - would think the same: not much going 
on here and no support given.
So why should he stay and start developing new desklets ?


To cut a long story short:
The main problem is NOT the lack of desklets and/or using ADL!
The problem is the way gDesklets is presented (or presenting itself) and
that it seems to be a bit "user unfriedly".

Easy to solve with not too much work to be done:

1. show activity
- update the homepage more frequently and keep users up to date, even if
there is not too much to talk about (include the release of new desklets
for example)
- include all working desklets from the archive site (even if it is just
one desklet a week)

2. make it much easier for new users to be helped
- set up a forum
- start a user wiki (or faq)
- ever thought about producing screencasts for the first basic steps
(like the installation tutorial)



OK, so much for that, Now finally another thought about the merge:

First of all I think gDesklets isn't outdated or anything like this. 
Sure you can do things differently and more "modern" etc. But a user 
won't be interested about things like this. He just wants nice looking 
and working desklets on his screen doing more or less useful things.
gDesklets might be a bit slow and big, and yes, the memory issus is a 
bad thing, but basically it is running relatively stable and OK (once it 
is running). That's all a user needs and is interested in.
The rest can still be done sooner or later, but "burying gDesklets" 
(more or less) now would be simply a bad idea.
And the whole thing sounds to me like you would like to bury gDesklets 
slowly!

> However the 0.4 branch is now lacking a lot of development manpower. 
> I feel that one big distraction is that it was a bit over designed in
>  the beginning.

Well... the more I think about it the merge looks to me like another
thing being overdesigned. Like heading for 0.5x before even starting
0.4x. gDesklets doesn't look ready for things like this to me right now.

> Instead of just doing widgets nicely, we wanted messaging over TCP,
> swappable rendering engines and window engines, different parsers,
> etc..

Ahhhhhh... there you go.
Maybe it is time to do exactly that NOW.
Maybe we should start doing widget nicely before thinking about a merge ?!

Just my 5 cents (as usual).

Greetings, Bjoern

PS: I really don't like to be a "I told you before" bothersome user, but 
most of the stuff has been on the mailing list a few month ago, even on 
one of the forums a few years ago...