Active Topics

 


Reply
Thread Tools
Posts: 3,074 | Thanked: 12,964 times | Joined on Mar 2010 @ Sofia,Bulgaria
#81
Originally Posted by Copernicus View Post
Ah! Here's some solid data, then. What would these errata be? What problems have they been causing, and how does using the CSSU fix them?
Judging from the way you've asked your question I am not sure you really want an answer. Anyway ARM errata 430973 is one of the two erratas which prevents thumb2 IS to be used on n900. And it has kernel fix. There is another errata re thumb2, which has its fix in binutils, but unfortunately not in the version of binutils in scratchbox. I am running thumb2 compiled Qt, gtk, hildon-desktop hildon-home and hildon-status-menu for more than 2 months on both devices I have, rock stable (as POC). Now, just those make 20MB difference in size compared to ARM compiled binaries. I assume you are experienced enough to draw your conclusions what that means.
 

The Following 7 Users Say Thank You to freemangordon For This Useful Post:
Copernicus's Avatar
Posts: 1,986 | Thanked: 7,698 times | Joined on Dec 2010 @ Dayton, Ohio
#82
Originally Posted by freemangordon View Post
Judging from the way you've asked your question I am not sure you really want an answer.
The anger with which I asked my question is due to the fact that nobody here seems to think the question is even worth asking, let alone answering. Give people wonderful bug fixes and they'll be happy, but treat them as know-nothing peons and they'll be pissed. So far, I've been getting a lot more of the know-nothing treatment and a lot less of the bug fix, thus my frustration.

But yeah, you've finally given me something I hadn't known before: the keywords "thumb2 IS". No, I'm not knowledgeable about the fine details of ARM architecture, but this helps. A quick Google, and I see that compiling with thumb2 does at least provides a nice reduction in code size. That is the first significant advantage to using the CSSU I've seen.

This needs to go on to the Wiki page! If nobody else does it, maybe I'll try to update it...
 

The Following 2 Users Say Thank You to Copernicus For This Useful Post:
Posts: 1,808 | Thanked: 4,272 times | Joined on Feb 2011 @ Germany
#83
Originally Posted by Copernicus View Post
That is the first significant advantage to using the CSSU I've seen.

This needs to go on to the Wiki page! If nobody else does it, maybe I'll try to update it...
Not so fast! AFAIK the CSSU does not include a kernel, so any kernel fixes plus recompilation of essential programs seems to be (for now) way beyond the scope of the CSSU.

But *I* would be interested in freemangordons-SU (I remove an "S" because I don't know how seamless that would be
 

The Following 2 Users Say Thank You to reinob For This Useful Post:
Copernicus's Avatar
Posts: 1,986 | Thanked: 7,698 times | Joined on Dec 2010 @ Dayton, Ohio
#84
Originally Posted by reinob View Post
AFAIK the CSSU does not include a kernel, so any kernel fixes plus recompilation of essential programs seems to be (for now) way beyond the scope of the CSSU.
Eh, I've kinda figured that the CSSU fixes and the various modified kernels would have to be taken together, as the folks working on the one seem to be the same ones working on the other, and so far as I can tell, there isn't much interest in keeping them separate from one another. (But who knows? There's practically no data available on what exactly the CSSU folks are doing, beyond statements like "fixes more than it breaks" or "go read the source code yourself". I do plan on reading the source code myself, eventually, though...)

But a potential 20% savings in code size? That, indeed, is something to lust after. I can see what I'm going to be spending my time researching this weekend.

Here's an interesting paper; they're only seeing a 10% improvement (over Thumb 1), but that's still nice:

http://www.cs.uiuc.edu/class/fa05/cs...ORS/Thumb2.pdf
 

The Following User Says Thank You to Copernicus For This Useful Post:
Mike Fila's Avatar
Posts: 412 | Thanked: 480 times | Joined on Feb 2011 @ Bronx, NY
#85
Originally Posted by Copernicus View Post
There's practically no data available on what exactly the CSSU folks are doing, beyond statements like "fixes more than it breaks" or "go read the source code yourself". I do plan on reading the source code myself, eventually, though...)

Oh course there is 477 pages worth of discussion leading to countless pages of the actual changes that were made. I say read the source code because that is where the actual documentation of what was done and when. Nevermind the collaboration that was done on IRC

http://gitorious.org/community-ssu

Is it feasible to list this in the wiki? No, the wiki is for the "end user" information on how to install and what new features are in the CSSU. CSSU is not a secret society anyone can be involved.

http://wiki.maemo.org/Community_SSU#Contributing

Again this is just a small group of devs and IMHO they did a damn fine job of trying to document what exactly was done so that people that want to see all the nuts and bolts of the project could.

I just want to point out the documentation for PR 1.3 is dismal in comparison. Keep this in mind that all the information needed to put together a project the size of CSSU was not available to the devs that made it happen.

http://talk.maemo.org/showthread.php?t=64343

Edit: The idea that they (CSSU team) just decided to package everything together for the hell of it is way off.

Last edited by Mike Fila; 2012-01-13 at 17:57.
 

The Following 5 Users Say Thank You to Mike Fila For This Useful Post:
Posts: 103 | Thanked: 211 times | Joined on Sep 2011
#86
As a developer, I would love to get CSSU on every N900 because of Qt 4.7.4 (kudos to freemangodon). Unfortunately, If I add a dependence on my software to Qt 4.7.4 features (mainly QtQuick 1.1 and an usable opengl graphics system), that work would never reach extras repository. :/
 

The Following 4 Users Say Thank You to inean For This Useful Post:
qwazix's Avatar
Moderator | Posts: 2,622 | Thanked: 5,447 times | Joined on Jan 2010
#87
Yes that its why I think CSSU must be pushed like other official updates. Then the user of course can ignore it.
Also There are many.applications that depend on kernel power like mobile hotspot
__________________
Proud coding competition 2012 winner: ρcam
My other apps: speedcrunch N9 N900 Jolla –– contactlaunch –– timenow

Nemo UX blog: Grog
My website: qwazix.com
My job: oob
 

The Following User Says Thank You to qwazix For This Useful Post:
Mike Fila's Avatar
Posts: 412 | Thanked: 480 times | Joined on Feb 2011 @ Bronx, NY
#88
Originally Posted by inean View Post
As a developer, I would love to get CSSU on every N900 because of Qt 4.7.4 (kudos to freemangodon). Unfortunately, If I add a dependence on my software to Qt 4.7.4 features (mainly QtQuick 1.1 and an usable opengl graphics system), that work would never reach extras repository. :/

Yep this is why I started this thread to find out why people weren't using it. After seeing what happened with cute tube I fear that marxian may stop updating it.

I'm sure your not the only dev that feels this way.
 

The Following User Says Thank You to Mike Fila For This Useful Post:
Copernicus's Avatar
Posts: 1,986 | Thanked: 7,698 times | Joined on Dec 2010 @ Dayton, Ohio
#89
Originally Posted by Mike Fila View Post
Oh course there is 477 pages worth of discussion leading to countless pages of the actual changes that were made. I say read the source code because that is where the actual documentation of what was done and when.
Cool! Then maybe in a few months or so, after I've cranked through 477 pages of discussion, I'll be in a position to know whether the CSSU is of value to me.

Is it feasible to list this in the wiki? No, the wiki is for the "end user" information on how to install and what new features are in the CSSU.
The whole point of devices like Wikipedia (or any encyclopedia) is to winnow down broad topics of information into bite-sized summaries that people can absorb quickly. A 477-page discussion is something that just screams for a good summarization, if it is of any value in the first place -- the cost of ascending that mountain of text is just so dang prohibitive. (And, if/when I finally do manage to get through that whole thing, I'll certainly create a summarization of it, if only for myself...)

Again this is just a small group of devs and IMHO they did a damn fine job of trying to document what exactly was done so that people that want to see all the nuts and bolts of the project could.
I personally believe that they've done quite a fine job of implementing the CSSU. If I can ever find the documentation, I'll see if they've done a good job of that as well. It's just going to take me some time to read through 477 pages of posts...

Edit: The idea that they (CSSU team) just decided to package everything together for the hell of it is way off.
I believe you! But, it'd be nice to have a little description somewhere about their packaging choices. (I imagine it is somewhere within the 477 pages of posts, so I'll find it eventually.)
 

The Following 2 Users Say Thank You to Copernicus For This Useful Post:
Posts: 3,074 | Thanked: 12,964 times | Joined on Mar 2010 @ Sofia,Bulgaria
#90
Originally Posted by inean View Post
As a developer, I would love to get CSSU on every N900 because of Qt 4.7.4 (kudos to freemangodon). Unfortunately, If I add a dependence on my software to Qt 4.7.4 features (mainly QtQuick 1.1 and an usable opengl graphics system), that work would never reach extras repository. :/
Hmm, isn't it possible to put Qt 474 in extras-devel? And use CSSU as upstream?

EDIT:
Scratch that, it will conflict with mp-fremantle-pr

Last edited by freemangordon; 2012-01-13 at 18:28.
 

The Following 2 Users Say Thank You to freemangordon For This Useful Post:
Reply


 
Forum Jump


All times are GMT. The time now is 09:51.