Xubuntu Strategy Document

[Eero Tamminen]

Hi,

On Sunday 25 May 2008, Vincent wrote:
> Cody announced it on his blog [1] but as I did not see it mentioned on
> this mailinglist I thought I'd mention it: the first draft of the Xubuntu
> Strategy Document [2] has been published and is ready for feedback.
> Cody's done an excellent job, but if you still feel the need to provide
> feedback or need more information otherwise, be sure to check out his
> blog post at [1].
>
> [1] http://cody.zapto.org/?p=25
> [2]
> https://wiki.ubuntu.com/Xubuntu/Specifications/Intrepid/StrategyDocument

Thanks!

On overall the documentation looks really good, but I have a few comments.


Focus 2: Usability -section:

This mentions just accessibility, but I would mention also localization.  If
a person doesn't understand English, the application is completely useless
for her if it's not localized for her language.  I think all the default
applications should be translated to the same languages that are supported
by the core Xfce, and this document (or release goals) should list these
languages.  If an application is missing more than (say) 1/4 of
a translation for one of these langauges, I would seriously consider
whether it should be included.


Focus 3: Performance -section:

"The initial target will be 128mbs-192mbs of RAM (with appropriate swap 
space available) and 333Mhz-400Mhz CPU."

I think this could say "Current minimum requirement" instead of
"The initial target" and I wonder how realistic this target is currently...
I would also leave swap out because it just makes testing harder
("Hm, the app finished after several hours after I added more swap,
is this now acceptable?").

This could also list a minimum set of test-cases that should be achievable 
with the given minimum system.  My proposal for the test-cases are:
- Installing Xubuntu from the live-CD
- After booting the default Xubuntu desktop, viewing the Xubuntu www-site
  with the default browser while playing music with the default music
  player and having default E-mail application and document editor
  open in  the background
With 2*RAM amount of swap, there could be also test-case of:
- Opening all default installed applications (not games nor settings)
  to their default state

It would be nice if there would be a link to a Wiki page that tells how to 
setup a virtual system for testing this minimal system configuration.


Incongruous Packages & Precidents -section:

* Packages that use interpreted languages are generally slower and require
   more memory then ones that are compiled.

-> Yes, they can be slower and take more memory, but they also take
     less disk space (if the required libraries are already installed). IMHO 
     decision should be based on the mission focus items and measurements.

* C is preferred over C++.

-> As Xubuntu's a distribution i.e. does packaging instead a coding, why
     this matters?  Because GCC C++ ABI hasn't so far been been very stable
     and there's been need for ABI transitions in Debian?

* Packages that will pull in a gnome lib that is not already pulled in can
   be not be seeded without the review and approval of the Xubuntu council.
* If possible, we should limit the number of gnome libs we pull in. [...]

-> I don't see why Gnome and libraries in general need to be singled out
    here.  They shouldn't be anymore "bad" than any other additional
    libraries.  The criteria should be based on relevant facts like
    increased (install ISO) disk size or RAM consumption instead of library
    names.

* Applications that use a lot of libs mean more memory usage - even more
  so for applications that use a lib exclusively. 

-> Again less useful blanket statement.  These kind of things should be
     based on facts & measurements.  If the application shares its
     functionality with other applications through libraries, in most cases
     that's disk & memory usage, security (better reviews & more testing),
     localization etc win compared to application suffering from NIH.

In general I think it's better to specify measurement test-cases, tools and
guideline limits for things to measure than above kind of guidelines.


Some typos in "Package Selection Matrix" section:
- "which doesn't makes it easy" -> "which makes it easy"
- "must not be taken likely" -> "lightly"
- "we should aim to reduce unrequired workload" -> "required workload"
- "Software is slow and takes up a lot of CPU? It might not"
  -> "Software that is slow and takes a lot of CPU might not"
- "does not provide it its self" -> "does not provide it by itself"?
- "software that is hosted on the Xfce4 server or a member of the xfce4
   goodies project"
   -> "software being hosted on the Xfce4 server or from a member of
         the xfc4 goodies project"
- "more important then if it is an (un)official Xfce4 side project"
  -> "more important than whether it's an"


	- Eero