Reply
Thread Tools
lcuk's Avatar
Posts: 1,635 | Thanked: 1,816 times | Joined on Apr 2008 @ Manchester, England
#131
b-man, install liqbase in native maemo
__________________
liqbase sketching the future.
like what i say? hit the Thanks, thanks!
twitter.com/lcuk
 
Posts: 3,841 | Thanked: 1,079 times | Joined on Nov 2006
#132
Re. that Python discussion.. I don't see any runtime performance issues with those apps I run. There seems to be a longer-than-normal startup time (compared to other applications). I suspect this is not entirely due to it being a scripting language (after all, command line scripts have no delay to talk of), so I suspect it's something to do with Python's UI elements. Which means that it could well be possible to fix that particular problem.

Re. left vs. right side of brain: The whole issue isn't particularly scientific, it's much more complicated (and balanced) than that. Left vs. right side brain usage is a discussion that mostly takes place in popular media. It doesn't mean it's true.

Edit: Here's even a link about it: http://en.wikipedia.org/wiki/Lateral...brain_function
__________________
N800/OS2007|N900/Maemo5
-- Metalayer-crawler delenda est.
-- Current state: Fed up with everything MeeGo.
 
Posts: 2,102 | Thanked: 1,309 times | Joined on Sep 2006
#133
Re. that Python discussion.. I don't see any runtime performance issues with those apps I run. There seems to be a longer-than-normal startup time (compared to other applications). I suspect this is not entirely due to it being a scripting language (after all, command line scripts have no delay to talk of), so I suspect it's something to do with Python's UI elements. Which means that it could well be possible to fix that particular problem.
I've noticed a delay on the first run of a program (the Python script is compiled to an intermediate "P-code" I think), after that it's fine.
 
pycage's Avatar
Posts: 3,404 | Thanked: 4,474 times | Joined on Oct 2005 @ Germany
#134
Slow code is slow, no matter in what language. Fast code is fast, no matter in what language. The difference between two languages is just a constant factor that becomes neglectable as soon as the application has to scale (e.g. handle millions of items instead of only a handful).
Unfortunately, fast computers nowadays give the illusion of fast although poorly written code, when ignoring scalation. That's why many people can't program well anymore, they just produce unscalable code.
Of course Python has some startup time. But why care as long as the application is responsive? I know enough C code stuff that is worse.
 

The Following User Says Thank You to pycage For This Useful Post:
lcuk's Avatar
Posts: 1,635 | Thanked: 1,816 times | Joined on Apr 2008 @ Manchester, England
#135
pycage,

I agree with this totally, and its why I always try to code for lowest denominator.

This device is quick if coded specifically for it and when used on other more powerful devices absolutely flies along (too fast actually, hundreds of fps in some cases!)

I managed to kill liqbase the other day by seeing what will happen when I have 25000+ sketches in my graffiti wall (currently I have ~1500), apart from running out of memory at the end (was consuming over 150mb!) the code was solid and scrolling and using the system was fluid and fast

note to self: implement a way to discard invisible stroke points when previewing to save memory
__________________
liqbase sketching the future.
like what i say? hit the Thanks, thanks!
twitter.com/lcuk
 
Posts: 2,152 | Thanked: 1,490 times | Joined on Jan 2006 @ Czech Republic
#136
Originally Posted by TA-t3 View Post
Re. that Python discussion.. I don't see any runtime performance issues with those apps I run. There seems to be a longer-than-normal startup time (compared to other applications). I suspect this is not entirely due to it being a scripting language (after all, command line scripts have no delay to talk of), so I suspect it's something to do with Python's UI elements. Which means that it could well be possible to fix that particular problem.
This is due to python gtk and hildon bindings, try to start python interpreter and then type 'import gtk'. Or make sample python code and print current time with millisecond precision after each import line on the beginning to see startup overhead of each module.

This is solved by optional python-launcher package in extras which keeps instance of already initialized python. This saves both time and memory when next python application is launched. See

http://www.mail-archive.com/maemo-de.../msg14875.html
https://garage.maemo.org/projects/pylauncher/
http://blogs.gnome.org/johan/2007/01...thon-launcher/
__________________
Newbies click here before posting. Thanks.

If you really need to PM me with troubleshooting question please consider posting it to the forum instead. It is OK to PM me a link to such post then. Thank you.
 

The Following 4 Users Say Thank You to fanoush For This Useful Post:
tso's Avatar
Posts: 4,783 | Thanked: 1,253 times | Joined on Aug 2007 @ norway
#137
do python launcher have much of an effect on battery life?
 
allnameswereout's Avatar
Posts: 3,397 | Thanked: 1,212 times | Joined on Jul 2008 @ Netherlands
#138
Benny, you're one of the few who has been able to take the time to write an in depth response without any flames. Thank you.

Originally Posted by benny1967 View Post
I can see the point of the simple "check mail while on the go" or "look up one site, then close browser" use cases.
Clearly, if this is what you have in mind, the current concept of Mameo isn't as ideal as the things you propose. (Hey! We agree here!)
Good. The point in this regard is also merely a default setting. IOW, assume the user wants an application fullscreen with a simple setting to make it not fullscreen (and more applications which make use of the powerful touchscreen). This is because this default setting is what new users (whom we want to attract to this environment) want.

It isn't as if you'll attract die-hard iPhone or Mac fans either. First you get the users inbetween Maemo and iPhone. Because the difference is so huge between the 2, there is a wide pool available. So whoever believes the next Maemo is going to be like an iPhone is a... well, she isn't realistic, nor is she able to discern my points to see the bigger picture.

I suspect these users simply don't want change, nor welcome new(, less tech-inclined) users. Perhaps the N8x0 is the last from Nokia they like (also given they appear to be the same ones not excited by the summit or the HS*PA chipset). I believe they should jump on the next hacker/niche platform.

But the point is: Would anybody pay $450,- or so for a device with only this limited capability? The things you talk about are things I do regularly on my S60 cell phone. Yes, this S60-phone could need a little love and attention in terms of UI design. It's not bad, but some of the points you make here about Maemo apply to S60. It could be improved.
If you take what I written in the paragraphs above into account I don't see how you can see much devaluation. You use the same Maemo environment, with some very powerful features in the core OS to make the device more user-friendly (these features became known on the Maemo summit). You get a HS*PA chipset which allows you 24/7 Internet connectivity. You do not have to use this chipset although indeed you do pay for it.

Regarding S60: indeed. And my wish is that Nokia brings the 2 platforms together so a user wouldn't even notice which OS is running under the hood. Or even much better: so that when the user wishes to, she could replace S60 with Linux or the other way around, running (almost) the same applications. But that is far fetched. Keep in mind S60 has from the ground, by default capability-based security features. Enabled.

[...]

Given these prices, I'd either want a Maemo-device that's a GSM-phone for €77,- or a Maemo-tablet that's not a phone for €45. If it is as restricted as you propose. If it's made for the casual "one task at a time" use case.

When something is in the €300-500 price range, it needs to do much, much more than just let me "open website, read, close". I want to do what I described in my first post: Things that require workflows across several applications.
Where have I suggested one is not allowed to run more than 1 application? I'm not suggesting one isn't able to copy/paste. There are simply tasks which do not require such option. If you are able to use Canola to use your NIT as a DAP you will not need an iPod. This makes the device more valuable. But you would want to run such application full screen. I believe applications like Canola, Carman, Liqbase add a huge value to the NIT because 1) they do one job a dedicted hardware device normally does very well 2) they provide a very user-friendly interface making use of the touchscreen (which you cannot on a nettop) 3) because they are user-friendly they attract new users who previously would buy a phone or nettop.

Some IRC channels are lame, you have several minutes between relevant messages. I use this time to go through my newsfeeds. You say the interaction between feed reader and browser is minimal? Not if many of the feeds you follow only provide headlines or teasers, but not the full text content. It's a constant back-and-forth between feed reader and browser. Then I find something I copy and send via mail right there an then, while I'm at it. I dont go through all the feeds and then close the feed reader and return to this one page and copy and then close the browser to open modest to ...
This is a usability issue though. Lets say you have your RSS reader convenienty in your browser, in full screen interface. After all, you are picking what you'd like to read more about. Once you click on a headline it shows the actual text of the article. You read it, and then you conveniently pick the next headline. There is no distraction. There are no flashing windows or back and forth between windows. You stay in the same interface. It might even be feasable to make the RSS reader visual (with the interface of the website as a thumbnail) as a SpeedDial.

I'm sorry to hear you consider my usage of my brain suboptimal. I feel I'm getting along just fine, maybe I'm lacking some of your social skills.
Some people just behave in a different way with a UI than other people. The way a programmer interacts is vastly different from most other people in a society.

What use would it be to make NITs for the same segment of the market that's already well covered by Apple? Wouldn't it be much wiser to go for the segment that's currently completely ignored by Apple? Not only would it reduce unnecessary fights, it also seems that Apple is holding the smaller segment (as compared to all the people who choose not to buy Apple, no matter which market) and the other one would be more profitable.
Because other corporations (including Nokia) are cloning the Apple paradigm to grab that big pool between the iPhone and the NIT, and I don't want to keep the NIT as a hacker device which only a few people use; I want to see Maemo, Linux and the NIT more widely used by people who also are able to use the device. For this the usability has to improve, as well as the UI. Else, if that doesn't happen, other corporations will make it happen and the NIT will eventually fade away like Sharp Zaurus.

Same here. I don't even remember people's names, let alone street names, yet I never get lost and find my way because I somehow always have a clear 3D model of where I am right now and how I got there from where I started. Nice chat. I just don't see how this relates to kiosk-mode applications that prevent task switching and adjusting volume etc.
Exactly. I mentioned it because it is a good example of how one uses visual imagination to complete a task. A graphics artist requires this.

I've been in the USA and I was shocked at the signs on the road. Very often the Americans use English text on their signs where we simply use symbols. If you'd just use symbols it'd be much easier. Yes, there is a learning curve to learn the meaning of the symbols, but once you know them it is very easy to follow.

Oh come on, you made this up right now.
Its just like how people only read the first page of search results on a search engine. They look, see if what they searched for is there, and then they continue or go back. If one would plan to go further than 1 page one would open a new tab. The back button is also used much more than the forward button which is why Mozilla makes the back button more prominent.

And even if they did: In MicroB, they do this by 1 click on a button that's always visible. In Fennec, they need 2 interactions, one of which is completely counter-intuitive. MicroB wins.
True, but once you are used to the Fennec interface you quickly tab 2 times to perform the gesture.

Among geeks, not in the real world. Geeks loved Operas mouse gestures. Nobody ever used them in real life. Geeks loved exotic skins on media players. 99,99% of the users wouldn't even find a play button with these skins.
Because there is a learning curve and because Opera is not very popular.

I'd argue that on a touch device gestures using touchscreen with finger are more user friendly than keybinds or clicking an icon especially if the gestures are equal in different applications (Apple is known for using the same binds in their whole platform).

The touchscreen is one of the components which makes the NIT unique; lets abuse it so much that it adds substantional value to the platform.

Ofcourse regular ports are still useful (I will also still use screen, ssh, and remote desktop), but not to most (new) users whom we wish to attract.
__________________
Goosfraba! All text written by allnameswereout is public domain unless stated otherwise. Thank you for sharing your output!
 
allnameswereout's Avatar
Posts: 3,397 | Thanked: 1,212 times | Joined on Jul 2008 @ Netherlands
#139
There's a very good reason why we still have the same menus, check-boxes and buttons we had decades ago: They stand the test of time. Every now and then, UI-gurus declare a new age of user interfaces. 20 years later, we happily click on buttons and choose from menus.
Backwards compatibility/legacy, relatively new market / user interface using touch, and hardware prices have gone down as well.

Yet, there are also advancements in user interfaces. Metadata is much more important. Windows Vista, MacOS X, and KDE/GNOME have various improvements in the user interfaces over the years.

Yet, there is also a lot of research in different user interfaces. BeOS did things different. Microsoft even researched e.g. with Singularity. Linux programmers as well, e.g. 3dwm, 3ddesktop, and later on Compiz. Virtual desktops is something Microsoft is starting to adopt, while Linux and Windows adopt Exposee-like task switcher.

I can't see how creating 653 profiles for all brightness/volume-combinations I could ever want would increase the joy of using the device. Still, even if I did: I'd need some means of activating a certain profile while I'm working in an application. And that was the whole point: You cannot do this if applications run full screen and full screen only.
Not 653 profiles. But my ancient Nokia phone even has profiles and allows me to create more. The settings are then in a profile, its saved for e.g. meeting, travelling, and so on. Why would my NIT not have such function? There are only a few combination of settings people normally use and that number is far closer near 0 than near 653.
__________________
Goosfraba! All text written by allnameswereout is public domain unless stated otherwise. Thank you for sharing your output!
 
allnameswereout's Avatar
Posts: 3,397 | Thanked: 1,212 times | Joined on Jul 2008 @ Netherlands
#140
Originally Posted by lardman View Post
I've noticed a delay on the first run of a program (the Python script is compiled to an intermediate "P-code" I think), after that it's fine.
Also, many C(++) libraries are already in use, while Python libraries aren't. For Maemo there is an application which preloads the Python libraries already so startup time is less. But this comes for a price, ofcourse.

There are also JITs available for scripting languages (JavaScript for example)

The compare is also moot on a different layer. When I developed LUA scripts to access a C library for SQL it was fast enough for deployment. This is similar to Python + QEdje. Its not as if QEdje is programmed in Python.

Compares by Rasterman between hardware X rendering and X.Org software rendering were both slower than software rendering by Evas/Edje.
__________________
Goosfraba! All text written by allnameswereout is public domain unless stated otherwise. Thank you for sharing your output!
 
Reply


 
Forum Jump


All times are GMT. The time now is 04:22.