Re: running a motherboard without a keyboard

[Guylhem Aznar]

A similar question about dcon.

>From dmesg, it looks just fine.
$ dmesg|grep dcon
[   26.481778] olpc-dcon: No DCON found on SMBus
[   30.651493] olpc-dcon:  Discovered DCON version 2
[   30.708411] PM: Adding info for platform:dcon

But apparently dcon sleep, output and freeze also refuse to change. Is
the lack of screen causing the problem, and how?
The brightness can be disabled however.

$ cat /sys/devices/platform/dcon/sleep
0
$ echo 1 > /sys/devices/platform/dcon/sleep
$ cat /sys/devices/platform/dcon/sleep
0
$ cat /sys/class/backlight/dcon-bl/brightness
15
$ echo 0 > /sys/class/backlight/dcon-bl/brightness
$ cat /sys/class/backlight/dcon-bl/brightness
0

The only guess I have is i2c_smbus_write_word_data , called as
dcon_write from  dcon_sleep is not working as it should.
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[Ed McNierney]

Yes.  That page was there, but was broken during a recent site update.  A
bug¹s been opened with our Web team and it will be fixed shortly.

- Ed


On 12/2/08 9:16 PM, "Bobby Powers" <[EMAIL PROTECTED]> wrote:

> 
> 
> On Tue, Dec 2, 2008 at 10:06 PM, Ed McNierney <[EMAIL PROTECTED]> wrote:
>> Yes, that's exactly the point.  We currently offer developer keys by
>> postal mail for users who do not have Internet access, but users
>> without Internet access cannot get the instructions on how to make
>> such a request.  The "Getting Started" pamphlet refers users to
>> www.laptop.org/source 
>>   for source code and mentions it is available on media, but without
>> Internet access you can't find out how to order the media.
> 
> if it refers to that link, are there plans to put some kind of page there?
> 
> bobby
>  
>> 
>> - Ed
>> 
>> P.S. And I see that even with Internet access, www.laptop.org/source
>> 
>> seems to not be there...
>> 
>> 
>> On Dec 2, 2008, at 3:47 PM, Chris Ball wrote:
>> 
>>> > Hi,
>>> >
 >> Yes.  For example, we can add information about postal-mail
 >> requests to the page, without opening up the 767 build.
>>> >
>>> > I think Ed's point is that someone who wants to take advantage of a
>>> > postal-mail request necessarily can't talk to activation.l.o.  :)
>>> >
>>> > - Chris.
>>> > --
>>> > Chris Ball   <[EMAIL PROTECTED]>
>> 
>> ___
>> Devel mailing list
>> Devel@lists.laptop.org
>> http://lists.laptop.org/listinfo/devel
> 
> 

___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[Bobby Powers]

On Tue, Dec 2, 2008 at 10:06 PM, Ed McNierney <[EMAIL PROTECTED]> wrote:

> Yes, that's exactly the point.  We currently offer developer keys by
> postal mail for users who do not have Internet access, but users
> without Internet access cannot get the instructions on how to make
> such a request.  The "Getting Started" pamphlet refers users to
> www.laptop.org/source
>  for source code and mentions it is available on media, but without
> Internet access you can't find out how to order the media.


if it refers to that link, are there plans to put some kind of page there?

bobby


>
>- Ed
>
> P.S. And I see that even with Internet access, www.laptop.org/source
> seems to not be there...
>
>
> On Dec 2, 2008, at 3:47 PM, Chris Ball wrote:
>
> > Hi,
> >
> >> Yes.  For example, we can add information about postal-mail
> >> requests to the page, without opening up the 767 build.
> >
> > I think Ed's point is that someone who wants to take advantage of a
> > postal-mail request necessarily can't talk to activation.l.o.  :)
> >
> > - Chris.
> > --
> > Chris Ball   <[EMAIL PROTECTED]>
>
> ___
> Devel mailing list
> Devel@lists.laptop.org
> http://lists.laptop.org/listinfo/devel
>
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: [Localization] [Proposal] .xot bundles, for translations

[Korakurider]

On Wed, Dec 3, 2008 at 3:47 AM, C. Scott Ananian <[EMAIL PROTECTED]> wrote:
> Re: Scratch & etoys:  the problem with updating translations "in
> place" is that it doesn't support distributed work on translations:
> OLPC might do basic translations; they might be further developed in a
> country or region, etc.  Each might be updated individually.
   Yes, current lang pack is being generated from translations on Pootle
   and can be used to replace older translations (that was from Pootle
also) in the box.
   Essential use case for the facility seems different from that what
you imagine (distributed translation work).

   I think that, even if some deployment want to do custom translation
in distributed fassion,
   they will need to build and deploy the authoritive version of
translation pack that they merge translations from their team and do
extensive human review.  So I believe lang pack will still work for
them.

>
> Further, you want to be able to back out changes, in order to protect
> against getting malicious "translations" from a friend.  Uninstalling
> a language bundle gives you that, un-merging changes to a shared
> in-place translation file is... more difficult.
Lang pack just replaces entire contents of each MO, not merge
contents from multiple MOs.  Sayamindu 's proposal is about how to
replace MO.  So uninstalling seems OK.

Sayamindu, could you confirm my understanding?

/Korakurider
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[Ed McNierney]

On Dec 2, 2008, at 6:52 PM, Richard A. Smith wrote:

If a thousand people do paper who is going to manually type in each  
of the key requests?


I thought that was your job.

The primary intent is to make developer keys available to users.  This  
is simply the same (rarely-used) process we advertise today at http://wiki.laptop.org/go/Developer_keys 
 as:


If you do not have internet access at all

This is very much not recommended, and extraordinarily rare, as most  
development work is done via email, chat, forums, and other internet- 
based communication mediums. However, if you must get a developer key  
and do not have access to the internet at all, you can submit a  
written request via snail mail to:


One Laptop per Child
P.O. Box 425087
Cambridge, MA 02142

Your key will be mailed back to you.

We are obligated to provide access to developer keys to children who  
have XO laptops but do not have Internet access.  We do have such a  
mechanism in place, but at the moment someone receiving an XO without  
Internet access can't get the information shown above.  Presumably  
only those people without access would want to do such a thing.


- Ed
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[Richard A. Smith]

Ed McNierney wrote:

> No, all I'm thinking about is having .devkey.html contain the postal  
> instructions in the HTML local to the machine, so it can be read by  
> someone without Internet access at all.  Someone can launch the Browse  
> activity, click on "get a developer key", and read the instructions  
> (and also, presumably, click the Submit button to make an online  
> request, too).

Request via postal mail?  Seriously? Is there some sort of license 
requirement for this?  Are they going to ship us a usb drive or sd card 
with the file leases.dat on it and then we ship back with the dev key on it?

Doing this via paper is absurd.  If you go through the support tickets 
you will find that we can't even get Joe non-tech person to type in the 
retrieval URL correctly much less a 256+ character string of what most 
would describe as gibberish.

If a thousand people do paper who is going to manually type in each of 
the key requests?

-- 
Richard Smith  <[EMAIL PROTECTED]>
One Laptop Per Child
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: [sugar] [Proposal] .xot bundles, for translations

[Martin Langhoff]

On Tue, Dec 2, 2008 at 7:34 PM, C. Scott Ananian <[EMAIL PROTECTED]> wrote:
> Please re-read Sayamindu's original message.  Thanks.

I don't find anything too special there. Perhaps I wasn't clear earlier.

What I meant to say is that all the good things we get from a bespoke
packaging format, we can get from rpm with a few conventions as to the
directories where things land.

You can still do "local user edits" by either editing the files
in-place, or editing a copy of the files, which is kept in a "local"
directory that is in the 'path' for localization files. The pros and
cons of both options can be argued separately, but both can work, and
many additional tricks can be put into action too.

So - I can't see anything that rpm can't handle, and I can't see an
interesting upside to building a bespoke mechanism to deploy files.
Perhaps it exists, and is really an overriding advantage that explains
why we'd want to carry the significant additional long term burden (on
us, and on everyone else) of a bespoke format.

cheers,



m
-- 
 [EMAIL PROTECTED]
 [EMAIL PROTECTED] -- School Server Architect
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

[Server-devel] XS 0.5 - eth1 not coming up

[Anna]

I installed XS 0.5 and am having trouble configuring the LAN.  I have 2 NICs
- eth0 to the outside world and eth1 facing the LAN.

I've read README.networking in /usr/share/doc/xs-config and looked at the
ifcfg scripts in /etc/sysconfig/network-scripts, but I still can't figure
out how landbond0 fits into the environment.  When I do ifconfig, it seems
like lanbond0 is getting assigned the stuff I'm used to seeing on eth1.
Eth1 doesn't get anything.

Here's the output of ifconfig for eth1 and the lanbond stuff:

eth1  Link encap:Ethernet  HWaddr 00:0D:56:05:7C:DA
  inet6 addr: fe80::20d:56ff:fe05:7cda/64 Scope:Link
  UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
  RX packets:0 errors:0 dropped:0 overruns:0 frame:0
  TX packets:6 errors:0 dropped:0 overruns:0 carrier:0
  collisions:0 txqueuelen:10
  RX bytes:0 (0.0 b)  TX bytes:468 (468.0 b)

lanbond0  Link encap:Ethernet  HWaddr 00:00:00:00:00:00
  inet addr:172.18.96.1  Bcast:172.18.127.255  Mask:255.255.224.0
  UP BROADCAST MASTER MULTICAST  MTU:1500  Metric:1
  RX packets:0 errors:0 dropped:0 overruns:0 frame:0
  TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
  collisions:0 txqueuelen:0
  RX bytes:0 (0.0 b)  TX bytes:0 (0.0 b)

lanbond0:1 Link encap:Ethernet  HWaddr 00:00:00:00:00:00
  inet addr:172.18.1.1  Bcast:172.18.1.255  Mask:255.255.255.0
  UP BROADCAST MASTER MULTICAST  MTU:1500  Metric:1

lanbond0:2 Link encap:Ethernet  HWaddr 00:00:00:00:00:00
  inet addr:172.18.0.1  Bcast:172.18.0.255  Mask:255.255.255.0
  UP BROADCAST MASTER MULTICAST  MTU:1500  Metric:1


Of course, being the adventuresome type, I did:

ifconfig eth1 172.18.96.0 netmask 255.255.224.0

ifdown lanbond0

Which broke everything till I rebooted.

Any suggestions on configuration?

Anna Schoolfield
Birmingham
___
Server-devel mailing list
[EMAIL PROTECTED]
http://lists.laptop.org/listinfo/server-devel

Re: Info on developer key request page on XO

[Ed McNierney]

No, all I'm thinking about is having .devkey.html contain the postal  
instructions in the HTML local to the machine, so it can be read by  
someone without Internet access at all.  Someone can launch the Browse  
activity, click on "get a developer key", and read the instructions  
(and also, presumably, click the Submit button to make an online  
request, too).

- Ed


On Dec 2, 2008, at 4:30 PM, C. Scott Ananian wrote:

> On Tue, Dec 2, 2008 at 4:06 PM, Ed McNierney <[EMAIL PROTECTED]> wrote:
>> Yes, that's exactly the point.  We currently offer developer keys  
>> by postal
>> mail for users who do not have Internet access, but users without  
>> Internet
>> access cannot get the instructions on how to make such a request.   
>> The
>> "Getting Started" pamphlet refers users to www.laptop.org/source  
>> for source
>> code and mentions it is available on media, but without Internet  
>> access you
>> can't find out how to order the media.
>
> Yes, I understand the point, but I don't see how this is worse than
> the alternative.  If I had hard-coded the activation lease info in the
> build, it *still* wouldn't have information about postal mail access,
> and there would be no way to put it there.
>
> For 9.1 we should have both: the library-common bundle is the right
> place to put information about dev keys; I hope that we will also
> replace the current form with a sugar control panel, which can have
> more information as well.
> --scott
>
> -- 
> ( http://cscott.net/ )

___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: [sugar] [Proposal] .xot bundles, for translations

[C. Scott Ananian]

On Tue, Dec 2, 2008 at 4:26 PM, Martin Langhoff
<[EMAIL PROTECTED]> wrote:
> On Tue, Dec 2, 2008 at 6:49 PM, C. Scott Ananian <[EMAIL PROTECTED]> wrote:
>> Fedora does not have a standard solution either,  so I'm not sure
>> where you're going with this.  We have to invent something.  RPM is
>> not obviously the right solution.
>
> So Fedora doesn't use rpm files for localization packages? What does
> it use then?
>
> If I say 'yum search catalan' it returns a bunch of rpms -
> kde-l10n-Catalan for example. What else could this mean?
>
> Debian does the same, AFAICS...

Please re-read Sayamindu's original message.  Thanks.
 --scott

-- 
 ( http://cscott.net/ )
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[C. Scott Ananian]

On Tue, Dec 2, 2008 at 4:06 PM, Ed McNierney <[EMAIL PROTECTED]> wrote:
> Yes, that's exactly the point.  We currently offer developer keys by postal
> mail for users who do not have Internet access, but users without Internet
> access cannot get the instructions on how to make such a request.  The
> "Getting Started" pamphlet refers users to www.laptop.org/source for source
> code and mentions it is available on media, but without Internet access you
> can't find out how to order the media.

Yes, I understand the point, but I don't see how this is worse than
the alternative.  If I had hard-coded the activation lease info in the
build, it *still* wouldn't have information about postal mail access,
and there would be no way to put it there.

For 9.1 we should have both: the library-common bundle is the right
place to put information about dev keys; I hope that we will also
replace the current form with a sugar control panel, which can have
more information as well.
 --scott

-- 
 ( http://cscott.net/ )
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: [sugar] [Proposal] .xot bundles, for translations

[Martin Langhoff]

On Tue, Dec 2, 2008 at 6:49 PM, C. Scott Ananian <[EMAIL PROTECTED]> wrote:
> Fedora does not have a standard solution either,  so I'm not sure
> where you're going with this.  We have to invent something.  RPM is
> not obviously the right solution.

So Fedora doesn't use rpm files for localization packages? What does
it use then?

If I say 'yum search catalan' it returns a bunch of rpms -
kde-l10n-Catalan for example. What else could this mean?

Debian does the same, AFAICS...

cheers,


m
-- 
 [EMAIL PROTECTED]
 [EMAIL PROTECTED] -- School Server Architect
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[Ed McNierney]

Yes, that's exactly the point.  We currently offer developer keys by  
postal mail for users who do not have Internet access, but users  
without Internet access cannot get the instructions on how to make  
such a request.  The "Getting Started" pamphlet refers users to 
www.laptop.org/source 
  for source code and mentions it is available on media, but without  
Internet access you can't find out how to order the media.

- Ed

P.S. And I see that even with Internet access, www.laptop.org/source  
seems to not be there...


On Dec 2, 2008, at 3:47 PM, Chris Ball wrote:

> Hi,
>
>> Yes.  For example, we can add information about postal-mail
>> requests to the page, without opening up the 767 build.
>
> I think Ed's point is that someone who wants to take advantage of a
> postal-mail request necessarily can't talk to activation.l.o.  :)
>
> - Chris.
> -- 
> Chris Ball   <[EMAIL PROTECTED]>

___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[Chris Ball]

Hi,

   > Yes.  For example, we can add information about postal-mail
   > requests to the page, without opening up the 767 build. 

I think Ed's point is that someone who wants to take advantage of a
postal-mail request necessarily can't talk to activation.l.o.  :)

- Chris.
-- 
Chris Ball   <[EMAIL PROTECTED]>
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: [sugar] [Proposal] .xot bundles, for translations

[C. Scott Ananian]

On Tue, Dec 2, 2008 at 2:26 PM, Martin Langhoff
<[EMAIL PROTECTED]> wrote:
> On Thu, Nov 13, 2008 at 6:58 PM, Sayamindu Dasgupta <[EMAIL PROTECTED]> wrote:
>> I have been thinking of having a separate place in the filesystem for
>> _new_ translations, and using RPM to manage the installation and
>> upgradation of the new translations.
>
> What is the downside of RPMs? If users edit the localisation locally,
> that is _fine_ and we can provide a mechanism to make an rpm easily
> out of it.
>
> rpm has limited support for "user installable" packages that are meant
> to be installed in your homedir. Maybe it can serve this purpose, even
> within its limitations?
>
> If that doesn't work properly, maybe we install the rpm as root, but
> invoking rpm with --noscripts, and perhaps auditing the pkg manifest
> to check for anything with suid flags, etc. We could even build a dumb
> rpm unpacker/installer but I doubt it is needed.
>
> A new bundle format makes us more incompatible with the world.
> Example:  someone builds a localisation for us, it won't work for
> Fedora, and viceversa.

Fedora does not have a standard solution either,  so I'm not sure
where you're going with this.  We have to invent something.  RPM is
not obviously the right solution.
 --scott


-- 
 ( http://cscott.net/ )
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[C. Scott Ananian]

On Tue, Dec 2, 2008 at 3:37 PM, Chris Ball <[EMAIL PROTECTED]> wrote:
> Hi Ed,
>
>   > Folks - I'm investigating steps that can make it easier to get
>   > developer keys for XOs for users who desire them.  At the moment
>   > the developer key request page at file:///home/.devkey.html
>   > contains content that is retrieved (in an IFRAME) from the
>   > activation.laptop.org server (as well as pulling a stylesheet from
>   > it).  Is there a particular reason for this approach as opposed to
>   > making some or all of that content static and local to the XO?
>
> My recollection of the reasoning is that we had a short time (before the
> original G1G1 image deadline) to pull this together, and decided that
> it's much more flexible to fix bugs and handle translations and giving
> instructions on devkey installation on the server-side, rather than
> needing to preload them all ahead of time.  Scott will know more, since
> he did the work.

Yes.  For example, we can add information about postal-mail requests
to the page, without opening up the 767 build.
 --scott

-- 
 ( http://cscott.net/ )
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Info on developer key request page on XO

[Chris Ball]

Hi Ed,

   > Folks - I'm investigating steps that can make it easier to get
   > developer keys for XOs for users who desire them.  At the moment
   > the developer key request page at file:///home/.devkey.html
   > contains content that is retrieved (in an IFRAME) from the
   > activation.laptop.org server (as well as pulling a stylesheet from
   > it).  Is there a particular reason for this approach as opposed to
   > making some or all of that content static and local to the XO?

My recollection of the reasoning is that we had a short time (before the
original G1G1 image deadline) to pull this together, and decided that
it's much more flexible to fix bugs and handle translations and giving
instructions on devkey installation on the server-side, rather than
needing to preload them all ahead of time.  Scott will know more, since
he did the work.

- Chris.
-- 
Chris Ball   <[EMAIL PROTECTED]>
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Info on developer key request page on XO

[Ed McNierney]

Folks -

I'm investigating steps that can make it easier to get developer keys  
for XOs for users who desire them.  At the moment the developer key  
request page at file:///home/.devkey.html contains content that is  
retrieved (in an IFRAME) from the activation.laptop.org server (as  
well as pulling a stylesheet from it).  Is there a particular reason  
for this approach as opposed to making some or all of that content  
static and local to the XO?

I'd like to have at least some of that content be local to the XO and  
contain instructions on how to request a developer key via postal  
mail.  The "make the entire page local" solution would seem to be an  
obvious approach, so I'm suspecting there was a reason to not do so.   
Comments, suggestions, and history are welcome - thanks!

- Ed
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: New joyride build 2570

[Daniel Drake]

On Mon, Dec 1, 2008 at 10:27 PM,  <[EMAIL PROTECTED]> wrote:
> is this ready for people to start testing yet?

Activity launching is broken, so I'd say this is still for developers
(probably for another month at least).

Daniel
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: [sugar] [Proposal] .xot bundles, for translations

[Martin Langhoff]

On Thu, Nov 13, 2008 at 6:58 PM, Sayamindu Dasgupta <[EMAIL PROTECTED]> wrote:
> I have been thinking of having a separate place in the filesystem for
> _new_ translations, and using RPM to manage the installation and
> upgradation of the new translations.

What is the downside of RPMs? If users edit the localisation locally,
that is _fine_ and we can provide a mechanism to make an rpm easily
out of it.

rpm has limited support for "user installable" packages that are meant
to be installed in your homedir. Maybe it can serve this purpose, even
within its limitations?

If that doesn't work properly, maybe we install the rpm as root, but
invoking rpm with --noscripts, and perhaps auditing the pkg manifest
to check for anything with suid flags, etc. We could even build a dumb
rpm unpacker/installer but I doubt it is needed.

A new bundle format makes us more incompatible with the world.
Example:  someone builds a localisation for us, it won't work for
Fedora, and viceversa.

Building bespoke sofware has a huge long term cost so when we do it,
we better get a ton of value, something radically better and hopefully
with immediate payoff.

Installing a bunch of files in /home is not leading edge enough to
justify this, IMHO.

cheers,


martin langhoff
-- 
 [EMAIL PROTECTED]
 [EMAIL PROTECTED] -- School Server Architect
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: [Localization] [Proposal] .xot bundles, for translations

[C. Scott Ananian]

Re: Scratch & etoys:  the problem with updating translations "in
place" is that it doesn't support distributed work on translations:
OLPC might do basic translations; they might be further developed in a
country or region, etc.  Each might be updated individually.

Further, you want to be able to back out changes, in order to protect
against getting malicious "translations" from a friend.  Uninstalling
a language bundle gives you that, un-merging changes to a shared
in-place translation file is... more difficult.

 --scott

-- 
 ( http://cscott.net/ )
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Journal2 recipe...

[C. Scott Ananian]

On Sat, Nov 29, 2008 at 11:13 PM, Michael Stone <[EMAIL PROTECTED]> wrote:
> Casual inspection of http://pinot.berlios.de/ suggested that the main
> 'pinot' GUI can be used to instruct the pinot-dbus-daemon to start
> indexing useful things like /home/olpc. In the future, it would be
> helpful if journal2 came with a sample ~/.pinot/config.xml or if its
> README instructed users to enable indexing in some other fashion.
>

Yes, pinot still needs to be configured using the standard GUI.
Patches welcome.
 --scott

-- 
 ( http://cscott.net/ )
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Journal2 patches + timings + notes

[C. Scott Ananian]

On Tue, Dec 2, 2008 at 1:27 AM, Michael Stone <[EMAIL PROTECTED]> wrote:
>  * startup time seems okay, but the initial query takes about 10s to
>return results. I have about 300 MB of data indexed under
>/home/olpc, plus a bit of other junk. (~/.pinot is ~ 35 MB).

Yes, that's because pinot is not actually returning incremental
results yet; it is doing the entire search, sorting it
chronologically, and *then* returning the first N.

>  * it takes around 4.2 seconds to fetch and display each new result

See above.  The subsequent searches have a hot cache, at least, but
they are not actually incremental.

>  * With today's code, the search system is effectively unusable while
>the indexer is hard at work.
>  * While I have not yet measured it carefully, it seemed to me that
>"full" indexing of a basic 767 system + activities _on XO_ may have
>taken as long as several hours.

See above; also the plan is to index /home, not /.

>  * With some, but not all queries, I noticed that the query failed
>because pinot failed to reply to journal2's blocking message sends
>within the usual dbus timeout of (30? 60?) seconds. I have no idea
>what was happening inside pinot.

See above.

> P.S. - Thanks for the stimulating vacation entertainment.

You're welcome!  Thanks for the patches!  Want to work some on pinot?
 --scott

-- 
 ( http://cscott.net/ )
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel

Re: Journal2 patches + timings + notes

[Martin Langhoff]

On Tue, Dec 2, 2008 at 4:27 AM, Michael Stone <[EMAIL PROTECTED]> wrote:
> I also did a bit of timing on my XO, (see the 'timing' branch), wherein
> I learned, _anecdotally_ (i.e.  with sample-size n=1) that:
>
>   * startup time seems okay, but the initial query takes about 10s to
> return results. I have about 300 MB of data indexed under
> /home/olpc, plus a bit of other junk. (~/.pinot is ~ 35 MB).
>   * it takes around 4.2 seconds to fetch and display each new result
>   * the standard deviation in my sample was large, e.g. about 3.3 seconds.
>   * 19 results were fetched.

What is the size of teh data indexed there? Working with git internals
I quickly realised that there are some good trivial strategies
(ab)using the fs performance on linux that are simpler and faster than
using an "indexing library" as I've done in the past.

On a slow fs like jffs2, fancy indexers can be even worse than the
cheapskates fs techniques because mmapping may suck on jffs2 but
(ab)using directories doesn't...

cheers,



m
-- 
 [EMAIL PROTECTED]
 [EMAIL PROTECTED] -- School Server Architect
 - ask interesting questions
 - don't get distracted with shiny stuff  - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff
___
Devel mailing list
Devel@lists.laptop.org
http://lists.laptop.org/listinfo/devel