squeaksource is down (again ...)

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
17 messages Options
Reply | Threaded
Open this post in threaded view
|

squeaksource is down (again ...)

laurent laffont
What can be done / how can we help / what's needed to improve squeaksource uptime ?

Laurent Laffont

http://pharocasts.blogspot.com/
http://magaloma.blogspot.com/

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Göran Krampe
On 06/24/2010 10:04 AM, laurent laffont wrote:
> What can be done / how can we help / what's needed to improve
> squeaksource uptime ?

I am not sure how SS is being deployed but one option might be to move
to another web server as a frontend and use FastCGI or SimpleCGI
(Blackfoot on SM).

My load tests a few years back on Nginx + Blackfoot compared to straight
KomHttpServer showed that the former combo handled load MUCH better, it
didn't miss requests.

Just an idea. Not sure exactly "why" though :)

regards, Göran

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Stéphane Ducasse

> I am not sure how SS is being deployed but one option might be to move to another web server as a frontend and use FastCGI or SimpleCGI (Blackfoot on SM).

could you give a try to build a new source with your solution?

>
> My load tests a few years back on Nginx + Blackfoot compared to straight KomHttpServer showed that the former combo handled load MUCH better, it didn't miss requests.
>
> Just an idea. Not sure exactly "why" though :)

For us we were thinking to move to a webdav server. But inria is concerned about security with free upload


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Philippe Marschall-2
In reply to this post by Göran Krampe
On 06/24/2010 12:17 PM, Göran Krampe wrote:
> On 06/24/2010 10:04 AM, laurent laffont wrote:
>> What can be done / how can we help / what's needed to improve
>> squeaksource uptime ?
>
> I am not sure how SS is being deployed but one option might be to move
> to another web server as a frontend and use FastCGI or SimpleCGI
> (Blackfoot on SM).

Believe me when I tell you, the web server is the least of the issues.
What mostly happens are Socket/Semaphore related issues, eg. a thousand
Processes hanging on the same Semaphore.

What could be done is moving to an up to date image/vm but this is a bit
of work.

Cheers
Philippe


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Lukas Renggli
I think it is the time to let squeaksource.com die. After all it
served its purpose and made Monticello what it is today: the standard
code and versioning system of Smalltalk fully supported by Pharo,
Squeak, and GemStone; and partially by Cincom Smalltalk and VA
Smalltalk.

I doubt that something can rescue squeaksource.com as it is now. Any
change means an incredible investment of work and the people that set
it up about 7 years ago are moving elsewhere.

I think somebody should jump in and provide a much better solution.

Lukas

On 24 June 2010 14:52, Philippe Marschall
<[hidden email]> wrote:

> On 06/24/2010 12:17 PM, Göran Krampe wrote:
>> On 06/24/2010 10:04 AM, laurent laffont wrote:
>>> What can be done / how can we help / what's needed to improve
>>> squeaksource uptime ?
>>
>> I am not sure how SS is being deployed but one option might be to move
>> to another web server as a frontend and use FastCGI or SimpleCGI
>> (Blackfoot on SM).
>
> Believe me when I tell you, the web server is the least of the issues.
> What mostly happens are Socket/Semaphore related issues, eg. a thousand
> Processes hanging on the same Semaphore.
>
> What could be done is moving to an up to date image/vm but this is a bit
> of work.
>
> Cheers
> Philippe
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project



--
Lukas Renggli
www.lukas-renggli.ch

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Richard Durr-2
What could such a solution be?

On Thu, Jun 24, 2010 at 7:52 PM, Lukas Renggli <[hidden email]> wrote:

> I think it is the time to let squeaksource.com die. After all it
> served its purpose and made Monticello what it is today: the standard
> code and versioning system of Smalltalk fully supported by Pharo,
> Squeak, and GemStone; and partially by Cincom Smalltalk and VA
> Smalltalk.
>
> I doubt that something can rescue squeaksource.com as it is now. Any
> change means an incredible investment of work and the people that set
> it up about 7 years ago are moving elsewhere.
>
> I think somebody should jump in and provide a much better solution.
>
> Lukas
>
> On 24 June 2010 14:52, Philippe Marschall
> <[hidden email]> wrote:
>> On 06/24/2010 12:17 PM, Göran Krampe wrote:
>>> On 06/24/2010 10:04 AM, laurent laffont wrote:
>>>> What can be done / how can we help / what's needed to improve
>>>> squeaksource uptime ?
>>>
>>> I am not sure how SS is being deployed but one option might be to move
>>> to another web server as a frontend and use FastCGI or SimpleCGI
>>> (Blackfoot on SM).
>>
>> Believe me when I tell you, the web server is the least of the issues.
>> What mostly happens are Socket/Semaphore related issues, eg. a thousand
>> Processes hanging on the same Semaphore.
>>
>> What could be done is moving to an up to date image/vm but this is a bit
>> of work.
>>
>> Cheers
>> Philippe
>>
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
>
>
> --
> Lukas Renggli
> www.lukas-renggli.ch
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Stéphane Ducasse
restart a nice application with a nice model.
Squeaksource was developed literally a couple of days. I remember when they said monday "wait until friday" :)

Stef

On Jun 25, 2010, at 12:33 AM, Richard Durr wrote:

> What could such a solution be?
>
> On Thu, Jun 24, 2010 at 7:52 PM, Lukas Renggli <[hidden email]> wrote:
>> I think it is the time to let squeaksource.com die. After all it
>> served its purpose and made Monticello what it is today: the standard
>> code and versioning system of Smalltalk fully supported by Pharo,
>> Squeak, and GemStone; and partially by Cincom Smalltalk and VA
>> Smalltalk.
>>
>> I doubt that something can rescue squeaksource.com as it is now. Any
>> change means an incredible investment of work and the people that set
>> it up about 7 years ago are moving elsewhere.
>>
>> I think somebody should jump in and provide a much better solution.
>>
>> Lukas
>>
>> On 24 June 2010 14:52, Philippe Marschall
>> <[hidden email]> wrote:
>>> On 06/24/2010 12:17 PM, Göran Krampe wrote:
>>>> On 06/24/2010 10:04 AM, laurent laffont wrote:
>>>>> What can be done / how can we help / what's needed to improve
>>>>> squeaksource uptime ?
>>>>
>>>> I am not sure how SS is being deployed but one option might be to move
>>>> to another web server as a frontend and use FastCGI or SimpleCGI
>>>> (Blackfoot on SM).
>>>
>>> Believe me when I tell you, the web server is the least of the issues.
>>> What mostly happens are Socket/Semaphore related issues, eg. a thousand
>>> Processes hanging on the same Semaphore.
>>>
>>> What could be done is moving to an up to date image/vm but this is a bit
>>> of work.
>>>
>>> Cheers
>>> Philippe
>>>
>>>
>>> _______________________________________________
>>> Pharo-project mailing list
>>> [hidden email]
>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>>
>>
>> --
>> Lukas Renggli
>> www.lukas-renggli.ch
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Geert Claes
Administrator
Maybe this is the time to re-visit/build upon SourceTalk ... but for Monticello 1?
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

laurent laffont
In reply to this post by Stéphane Ducasse
On Fri, Jun 25, 2010 at 8:21 AM, Stéphane Ducasse <[hidden email]> wrote:
restart a nice application with a nice model.
Squeaksource was developed literally a couple of days. I remember when they said monday "wait until friday" :)

What should be in ? I'm thinking of:

Important:
- fork / merge tracking (like github, so we can contribute to a package even if we don't have write access on main repository)
- scriptable API (automate patch integration from PharoInbox, image building)
- easy mirroring 

Would be nice:
- documentation integration (automatically display HelpSystem doc. on web page)
- issues tracking + project management 

 

Stef

On Jun 25, 2010, at 12:33 AM, Richard Durr wrote:

> What could such a solution be?
>
> On Thu, Jun 24, 2010 at 7:52 PM, Lukas Renggli <[hidden email]> wrote:
>> I think it is the time to let squeaksource.com die. After all it
>> served its purpose and made Monticello what it is today: the standard
>> code and versioning system of Smalltalk fully supported by Pharo,
>> Squeak, and GemStone; and partially by Cincom Smalltalk and VA
>> Smalltalk.
>>
>> I doubt that something can rescue squeaksource.com as it is now. Any
>> change means an incredible investment of work and the people that set
>> it up about 7 years ago are moving elsewhere.
>>
>> I think somebody should jump in and provide a much better solution.
>>
>> Lukas
>>
>> On 24 June 2010 14:52, Philippe Marschall
>> <[hidden email]> wrote:
>>> On 06/24/2010 12:17 PM, Göran Krampe wrote:
>>>> On 06/24/2010 10:04 AM, laurent laffont wrote:
>>>>> What can be done / how can we help / what's needed to improve
>>>>> squeaksource uptime ?
>>>>
>>>> I am not sure how SS is being deployed but one option might be to move
>>>> to another web server as a frontend and use FastCGI or SimpleCGI
>>>> (Blackfoot on SM).
>>>
>>> Believe me when I tell you, the web server is the least of the issues.
>>> What mostly happens are Socket/Semaphore related issues, eg. a thousand
>>> Processes hanging on the same Semaphore.
>>>
>>> What could be done is moving to an up to date image/vm but this is a bit
>>> of work.
>>>
>>> Cheers
>>> Philippe
>>>
>>>
>>> _______________________________________________
>>> Pharo-project mailing list
>>> [hidden email]
>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>>
>>
>> --
>> Lukas Renggli
>> www.lukas-renggli.ch
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Geert Claes
Administrator
Feature wise, I don't think SqueakSource's "Groups" is very useful, but I agree with the suggestion that it would be nice to have an integrated Issue Tracking (maybe Gjallar?) system (even an integrated Wiki/CMS).  Looking at systems like Trac, it has proven to very successful.

Looking at other online source code repositories is probably a good idea.  BitBucket for example doens't use the term "project" but actually let's you add a "Repository" which terminology wise blends better with Pharo as well.

I am happy to help out a bit from a usability design point of view.
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Geert Claes
Administrator
In reply to this post by laurent laffont
Down again now...

Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

laurent laffont


On Fri, Jun 25, 2010 at 9:50 AM, Geert Claes <[hidden email]> wrote:

Down again now...
http://forum.world.st/file/n2268082/Squeaksource_down_again.png


Maybe the good news is that Pharo / Squeak is more popular ;) 


Laurent

 


--
View this message in context: http://forum.world.st/squeaksource-is-down-again-tp2266603p2268082.html
Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Geert Claes
Administrator
laurent laffont wrote
Maybe the good news is that Pharo / Squeak is more popular ;)
A true optimist you are Laurent, good on ya mate :)
Reply | Threaded
Open this post in threaded view
|

Re: [SPAM] Re: squeaksource is down (again ...)

Sven Van Caekenberghe
In reply to this post by Geert Claes

On 25 Jun 2010, at 09:50, Geert Claes wrote:

>
> Down again now...
> http://forum.world.st/file/n2268082/Squeaksource_down_again.png 
>
> --
> View this message in context: http://forum.world.st/squeaksource-is-down-again-tp2266603p2268082.html
> Sent from the Pharo Smalltalk mailing list archive at Nabble.com.
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Actually the core protocol between an MC client and an MC server is dead simple,
here is an implementation working in Pharo 1.1

http://forum.world.st/WebClient-Monticello-tp2264855p2264855.html

But the full SqueakSource functionality is a different beast I think.

Sven


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Lukas Renggli
In reply to this post by Geert Claes
> Maybe this is the time to re-visit/build upon SourceTalk ... but for
> Monticello 1?

The problem is that SourceTalk is quite heavily targeted at Monticello
2. For example, projects directly map to Monticello 2 projects. Also I
suspect that it would also require some additional tweaks to really
make it scale better than SqueakSource in the next 10 years.

I am still waiting for Monticello 2 to finally come out. It would be
so much a step forward. And we would get rid of the legacy
SqueakSource code.

Lukas

> --
> View this message in context: http://forum.world.st/squeaksource-is-down-again-tp2266603p2268039.html
> Sent from the Pharo Smalltalk mailing list archive at Nabble.com.
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>



--
Lukas Renggli
www.lukas-renggli.ch

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: [SPAM] Re: squeaksource is down (again ...)

laurent laffont
In reply to this post by Sven Van Caekenberghe


Actually the core protocol between an MC client and an MC server is dead simple,
here is an implementation working in Pharo 1.1

http://forum.world.st/WebClient-Monticello-tp2264855p2264855.html


So cool :)
 
But the full SqueakSource functionality is a different beast I think.

Analyzing mcz to show branching / fork using ancestors may not be difficult.

I've just remembered this:

Laurent
 

Sven


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
Reply | Threaded
Open this post in threaded view
|

Re: squeaksource is down (again ...)

Geert Claes
Administrator
In reply to this post by Lukas Renggli
Lukas Renggli wrote
The problem is that SourceTalk is quite heavily targeted at Monticello 2. For example, projects directly map to Monticello 2 projects. Also I suspect that it would also require some additional tweaks to really make it scale better than SqueakSource in the next 10 years.

I am still waiting for Monticello 2 to finally come out. It would be so much a step forward. And we would get rid of the legacy SqueakSource code.
Sounds like a strategic decision maybe required?  When is MC2 scheduled (or expected) to come out?  Maybe a Pharo MC2 sprint could speed things up?

ps. I have to say, I probably just really like the name SourceTalk ... it's a neutral name for GemStone, Pharo, Squeak ... and could even have a domain like "st.world.st" :)