An on-going concern of the community is sustaining and improving our
trunk development process. Currently source.squeak.org employs an old version of SqueakSource running on an old Squeak 3.x image on an old interpreter VM. I have deployed a prototype replacement on box4 with several improvements: - It is running on the latest trunk image with a recent Cog VM. - The SqueakSource package has been upgraded to generically support any type of backend instead of only a Filesystem-based back end. - A new subclass for a Magma-based back end was added and used by the prototype. - Startup time is > 10X faster than with the FileSystem (which uses ReferenceStream). - Commits are > 100X faster because it's simply a DB commit rather than re-serializing the entire model with ReferenceStream for every change. - However, we will not notice a speed up because we have hacked SqueakSource to bypass the Filesystem subclass for a save-the-image-every-hour persistence. But this is unsustainable and dangerous. - The SqueakSource-software now responds to SIGTERM to conduct a graceful shutdown and exit. - Use of the Magma back end improves the SCM tool by letting us browse all 'mc versions' of any method or class-definition. We can also browse the 'mc origin' of same -- which is just the oldest version it is known to exist in (which, if we would ensure ALL ancestry is present, then it will truly be the origin). Happy birthday Tim Rowledge. I have done VERY minimal testing. But we need to finally begin the process of moving this crucial component our community away from the costly Hetzner box and onto Gandi-based servers provided by the SFC, so please try it and provide feedback! To use the new prototype: 1) Use a recent trunk image. 2) Install Monticello-cmm.565 from the Inbox. 'browse mc versions' and 'browse mc origin' will be added to methods and class-list menus (but are initially grayed out). 3) Add this repository to Monticello: MCHttpRepository location: 'http://box4.squeak.org:8888/trunk' user: 'id' password: 'pw' 4) Add the above repository to whichever packages you wish to invoke the new SCM features. - Chris PS -- This prototype only has data loaded for the trunk project as of about one week ago. It is not kept in-sync with production trunk, this is just a test. |
> 2) Install Monticello-cmm.565 from the Inbox.
Monticello-cmm.566 |
In reply to this post by Chris Muller-4
Bravo! This is great stuff.
I loaded Monticello-cmm.566 from the inbox, and I want to try out the new 'browse mc origin' and 'browse mc versions' feature, which shows up in the methods pane and class list pane as promised. But I can't figure out how to activate them so they are not greyed out??? I'm trying to browse a class in the sound package (just to see what it does), so I added the http://box4.squeak.org:8888/trunk repository to package Sound in my MC browser, but no joy. How do I activate the 'browse mc versions' feature? Thanks for doing this Chris! Dave On Fri, Sep 06, 2013 at 04:46:46PM -0500, Chris Muller wrote: > An on-going concern of the community is sustaining and improving our > trunk development process. Currently source.squeak.org employs an old > version of SqueakSource running on an old Squeak 3.x image on an old > interpreter VM. I have deployed a prototype replacement on box4 with > several improvements: > > - It is running on the latest trunk image with a recent Cog VM. > - The SqueakSource package has been upgraded to generically support > any type of backend instead of only a Filesystem-based back end. > - A new subclass for a Magma-based back end was added and used by > the prototype. > - Startup time is > 10X faster than with the FileSystem (which uses > ReferenceStream). > - Commits are > 100X faster because it's simply a DB commit rather > than re-serializing the entire model with ReferenceStream for every > change. > - However, we will not notice a speed up because we have hacked > SqueakSource to bypass the Filesystem subclass for a > save-the-image-every-hour persistence. But this is unsustainable and > dangerous. > - The SqueakSource-software now responds to SIGTERM to conduct a > graceful shutdown and exit. > - Use of the Magma back end improves the SCM tool by letting us > browse all 'mc versions' of any method or class-definition. We can > also browse the 'mc origin' of same -- which is just the oldest > version it is known to exist in (which, if we would ensure ALL > ancestry is present, then it will truly be the origin). Happy > birthday Tim Rowledge. > > I have done VERY minimal testing. But we need to finally begin the > process of moving this crucial component our community away from the > costly Hetzner box and onto Gandi-based servers provided by the SFC, > so please try it and provide feedback! > > To use the new prototype: > > 1) Use a recent trunk image. > 2) Install Monticello-cmm.565 from the Inbox. 'browse mc versions' > and 'browse mc origin' will be added to methods and class-list menus > (but are initially grayed out). > 3) Add this repository to Monticello: > > MCHttpRepository > location: 'http://box4.squeak.org:8888/trunk' > user: 'id' > password: 'pw' > > 4) Add the above repository to whichever packages you wish to invoke > the new SCM features. > > - Chris > > PS -- This prototype only has data loaded for the trunk project as of > about one week ago. It is not kept in-sync with production trunk, > this is just a test. |
On 06-09-2013, at 4:22 PM, "David T. Lewis" <[hidden email]> wrote: > Bravo! This is great stuff. Seconded. I'm stuck deep in the bowels of the old image right now so I can't sensibly try this out yet but I'm looking forward to it. tim -- tim Rowledge; [hidden email]; http://www.rowledge.org/tim Strange OpCodes: JSP: Jump on Sexy Programmer |
In reply to this post by David T. Lewis
It looks like I've missed a few methods to support those new functions
that need to be moved to Monticello (from Magma). Sorry, I'll post a fix this weekend.. On Fri, Sep 6, 2013 at 6:22 PM, David T. Lewis <[hidden email]> wrote: > Bravo! This is great stuff. > > I loaded Monticello-cmm.566 from the inbox, and I want to try out the > new 'browse mc origin' and 'browse mc versions' feature, which shows > up in the methods pane and class list pane as promised. But I can't > figure out how to activate them so they are not greyed out??? I'm trying > to browse a class in the sound package (just to see what it does), so > I added the http://box4.squeak.org:8888/trunk repository to package > Sound in my MC browser, but no joy. How do I activate the 'browse mc versions' > feature? > > Thanks for doing this Chris! > > Dave > > On Fri, Sep 06, 2013 at 04:46:46PM -0500, Chris Muller wrote: >> An on-going concern of the community is sustaining and improving our >> trunk development process. Currently source.squeak.org employs an old >> version of SqueakSource running on an old Squeak 3.x image on an old >> interpreter VM. I have deployed a prototype replacement on box4 with >> several improvements: >> >> - It is running on the latest trunk image with a recent Cog VM. >> - The SqueakSource package has been upgraded to generically support >> any type of backend instead of only a Filesystem-based back end. >> - A new subclass for a Magma-based back end was added and used by >> the prototype. >> - Startup time is > 10X faster than with the FileSystem (which uses >> ReferenceStream). >> - Commits are > 100X faster because it's simply a DB commit rather >> than re-serializing the entire model with ReferenceStream for every >> change. >> - However, we will not notice a speed up because we have hacked >> SqueakSource to bypass the Filesystem subclass for a >> save-the-image-every-hour persistence. But this is unsustainable and >> dangerous. >> - The SqueakSource-software now responds to SIGTERM to conduct a >> graceful shutdown and exit. >> - Use of the Magma back end improves the SCM tool by letting us >> browse all 'mc versions' of any method or class-definition. We can >> also browse the 'mc origin' of same -- which is just the oldest >> version it is known to exist in (which, if we would ensure ALL >> ancestry is present, then it will truly be the origin). Happy >> birthday Tim Rowledge. >> >> I have done VERY minimal testing. But we need to finally begin the >> process of moving this crucial component our community away from the >> costly Hetzner box and onto Gandi-based servers provided by the SFC, >> so please try it and provide feedback! >> >> To use the new prototype: >> >> 1) Use a recent trunk image. >> 2) Install Monticello-cmm.565 from the Inbox. 'browse mc versions' >> and 'browse mc origin' will be added to methods and class-list menus >> (but are initially grayed out). >> 3) Add this repository to Monticello: >> >> MCHttpRepository >> location: 'http://box4.squeak.org:8888/trunk' >> user: 'id' >> password: 'pw' >> >> 4) Add the above repository to whichever packages you wish to invoke >> the new SCM features. >> >> - Chris >> >> PS -- This prototype only has data loaded for the trunk project as of >> about one week ago. It is not kept in-sync with production trunk, >> this is just a test. > |
In reply to this post by Chris Muller-4
On 9/6/13 6:46 PM, "Chris Muller" <[hidden email]> wrote: > An on-going concern of the community is sustaining and improving our > trunk development process. Currently source.squeak.org employs an old > version of SqueakSource running on an old Squeak 3.x image on an old > interpreter VM. I have deployed a prototype replacement on box4 with > several improvements: > > - It is running on the latest trunk image with a recent Cog VM. > - The SqueakSource package has been upgraded to generically support > any type of backend instead of only a Filesystem-based back end. > - A new subclass for a Magma-based back end was added and used by > the prototype. > - Startup time is > 10X faster than with the FileSystem (which uses > ReferenceStream). > - Commits are > 100X faster because it's simply a DB commit rather > than re-serializing the entire model with ReferenceStream for every > change. > - However, we will not notice a speed up because we have hacked > SqueakSource to bypass the Filesystem subclass for a > save-the-image-every-hour persistence. But this is unsustainable and > dangerous. > - The SqueakSource-software now responds to SIGTERM to conduct a > graceful shutdown and exit. > - Use of the Magma back end improves the SCM tool by letting us > browse all 'mc versions' of any method or class-definition. We can > also browse the 'mc origin' of same -- which is just the oldest > version it is known to exist in (which, if we would ensure ALL > ancestry is present, then it will truly be the origin). Happy > birthday Tim Rowledge. > > I have done VERY minimal testing. But we need to finally begin the > process of moving this crucial component our community away from the > costly Hetzner box and onto Gandi-based servers provided by the SFC, > so please try it and provide feedback! > > To use the new prototype: > > 1) Use a recent trunk image. > 2) Install Monticello-cmm.565 from the Inbox. 'browse mc versions' > and 'browse mc origin' will be added to methods and class-list menus > (but are initially grayed out). > 3) Add this repository to Monticello: > > MCHttpRepository > location: 'http://box4.squeak.org:8888/trunk' > user: 'id' > password: 'pw' > > 4) Add the above repository to whichever packages you wish to invoke > the new SCM features. > > - Chris > > PS -- This prototype only has data loaded for the trunk project as of > about one week ago. It is not kept in-sync with production trunk, > this is just a test. > BRAVO!! I like to know if the .image could be downloaded from some place for learn of it. Edgar |
In reply to this post by David T. Lewis
Ok, please merge Monticello-cmm.568 from the Inbox and the new SCM
features should work! On Fri, Sep 6, 2013 at 6:22 PM, David T. Lewis <[hidden email]> wrote: > Bravo! This is great stuff. > > I loaded Monticello-cmm.566 from the inbox, and I want to try out the > new 'browse mc origin' and 'browse mc versions' feature, which shows > up in the methods pane and class list pane as promised. But I can't > figure out how to activate them so they are not greyed out??? I'm trying > to browse a class in the sound package (just to see what it does), so > I added the http://box4.squeak.org:8888/trunk repository to package > Sound in my MC browser, but no joy. How do I activate the 'browse mc versions' > feature? > > Thanks for doing this Chris! > > Dave > > On Fri, Sep 06, 2013 at 04:46:46PM -0500, Chris Muller wrote: >> An on-going concern of the community is sustaining and improving our >> trunk development process. Currently source.squeak.org employs an old >> version of SqueakSource running on an old Squeak 3.x image on an old >> interpreter VM. I have deployed a prototype replacement on box4 with >> several improvements: >> >> - It is running on the latest trunk image with a recent Cog VM. >> - The SqueakSource package has been upgraded to generically support >> any type of backend instead of only a Filesystem-based back end. >> - A new subclass for a Magma-based back end was added and used by >> the prototype. >> - Startup time is > 10X faster than with the FileSystem (which uses >> ReferenceStream). >> - Commits are > 100X faster because it's simply a DB commit rather >> than re-serializing the entire model with ReferenceStream for every >> change. >> - However, we will not notice a speed up because we have hacked >> SqueakSource to bypass the Filesystem subclass for a >> save-the-image-every-hour persistence. But this is unsustainable and >> dangerous. >> - The SqueakSource-software now responds to SIGTERM to conduct a >> graceful shutdown and exit. >> - Use of the Magma back end improves the SCM tool by letting us >> browse all 'mc versions' of any method or class-definition. We can >> also browse the 'mc origin' of same -- which is just the oldest >> version it is known to exist in (which, if we would ensure ALL >> ancestry is present, then it will truly be the origin). Happy >> birthday Tim Rowledge. >> >> I have done VERY minimal testing. But we need to finally begin the >> process of moving this crucial component our community away from the >> costly Hetzner box and onto Gandi-based servers provided by the SFC, >> so please try it and provide feedback! >> >> To use the new prototype: >> >> 1) Use a recent trunk image. >> 2) Install Monticello-cmm.565 from the Inbox. 'browse mc versions' >> and 'browse mc origin' will be added to methods and class-list menus >> (but are initially grayed out). >> 3) Add this repository to Monticello: >> >> MCHttpRepository >> location: 'http://box4.squeak.org:8888/trunk' >> user: 'id' >> password: 'pw' >> >> 4) Add the above repository to whichever packages you wish to invoke >> the new SCM features. >> >> - Chris >> >> PS -- This prototype only has data loaded for the trunk project as of >> about one week ago. It is not kept in-sync with production trunk, >> this is just a test. > |
In reply to this post by Edgar De Cleene
It's in my home directory on box4, but probably the best way to
understand what I did is to diff SqueakSource-cmm.1077 with SqueakSource-bf.1074 versions in http://source.squeak.org/ss. On Sat, Sep 7, 2013 at 9:56 AM, Edgar J. De Cleene <[hidden email]> wrote: > > > > On 9/6/13 6:46 PM, "Chris Muller" <[hidden email]> wrote: > >> An on-going concern of the community is sustaining and improving our >> trunk development process. Currently source.squeak.org employs an old >> version of SqueakSource running on an old Squeak 3.x image on an old >> interpreter VM. I have deployed a prototype replacement on box4 with >> several improvements: >> >> - It is running on the latest trunk image with a recent Cog VM. >> - The SqueakSource package has been upgraded to generically support >> any type of backend instead of only a Filesystem-based back end. >> - A new subclass for a Magma-based back end was added and used by >> the prototype. >> - Startup time is > 10X faster than with the FileSystem (which uses >> ReferenceStream). >> - Commits are > 100X faster because it's simply a DB commit rather >> than re-serializing the entire model with ReferenceStream for every >> change. >> - However, we will not notice a speed up because we have hacked >> SqueakSource to bypass the Filesystem subclass for a >> save-the-image-every-hour persistence. But this is unsustainable and >> dangerous. >> - The SqueakSource-software now responds to SIGTERM to conduct a >> graceful shutdown and exit. >> - Use of the Magma back end improves the SCM tool by letting us >> browse all 'mc versions' of any method or class-definition. We can >> also browse the 'mc origin' of same -- which is just the oldest >> version it is known to exist in (which, if we would ensure ALL >> ancestry is present, then it will truly be the origin). Happy >> birthday Tim Rowledge. >> >> I have done VERY minimal testing. But we need to finally begin the >> process of moving this crucial component our community away from the >> costly Hetzner box and onto Gandi-based servers provided by the SFC, >> so please try it and provide feedback! >> >> To use the new prototype: >> >> 1) Use a recent trunk image. >> 2) Install Monticello-cmm.565 from the Inbox. 'browse mc versions' >> and 'browse mc origin' will be added to methods and class-list menus >> (but are initially grayed out). >> 3) Add this repository to Monticello: >> >> MCHttpRepository >> location: 'http://box4.squeak.org:8888/trunk' >> user: 'id' >> password: 'pw' >> >> 4) Add the above repository to whichever packages you wish to invoke >> the new SCM features. >> >> - Chris >> >> PS -- This prototype only has data loaded for the trunk project as of >> about one week ago. It is not kept in-sync with production trunk, >> this is just a test. >> > > BRAVO!! > > I like to know if the .image could be downloaded from some place for learn > of it. > > Edgar > > > |
In reply to this post by Chris Muller-3
With Monticello-cmm.569 it seems to be doing the right things. Very nice!
One question - Browse mc versions gives a patch browser that displays the versions, but it does not seem to be showing method timestamp and author initial information, or the mc origin of each method version. Is this not yet implemented, or am I missing something in my image that is present in yours? Thanks, Dave On Sat, Sep 07, 2013 at 11:33:15AM -0500, Chris Muller wrote: > Ok, please merge Monticello-cmm.568 from the Inbox and the new SCM > features should work! > > On Fri, Sep 6, 2013 at 6:22 PM, David T. Lewis <[hidden email]> wrote: > > Bravo! This is great stuff. > > > > I loaded Monticello-cmm.566 from the inbox, and I want to try out the > > new 'browse mc origin' and 'browse mc versions' feature, which shows > > up in the methods pane and class list pane as promised. But I can't > > figure out how to activate them so they are not greyed out??? I'm trying > > to browse a class in the sound package (just to see what it does), so > > I added the http://box4.squeak.org:8888/trunk repository to package > > Sound in my MC browser, but no joy. How do I activate the 'browse mc versions' > > feature? > > > > Thanks for doing this Chris! > > > > Dave > > > > On Fri, Sep 06, 2013 at 04:46:46PM -0500, Chris Muller wrote: > >> An on-going concern of the community is sustaining and improving our > >> trunk development process. Currently source.squeak.org employs an old > >> version of SqueakSource running on an old Squeak 3.x image on an old > >> interpreter VM. I have deployed a prototype replacement on box4 with > >> several improvements: > >> > >> - It is running on the latest trunk image with a recent Cog VM. > >> - The SqueakSource package has been upgraded to generically support > >> any type of backend instead of only a Filesystem-based back end. > >> - A new subclass for a Magma-based back end was added and used by > >> the prototype. > >> - Startup time is > 10X faster than with the FileSystem (which uses > >> ReferenceStream). > >> - Commits are > 100X faster because it's simply a DB commit rather > >> than re-serializing the entire model with ReferenceStream for every > >> change. > >> - However, we will not notice a speed up because we have hacked > >> SqueakSource to bypass the Filesystem subclass for a > >> save-the-image-every-hour persistence. But this is unsustainable and > >> dangerous. > >> - The SqueakSource-software now responds to SIGTERM to conduct a > >> graceful shutdown and exit. > >> - Use of the Magma back end improves the SCM tool by letting us > >> browse all 'mc versions' of any method or class-definition. We can > >> also browse the 'mc origin' of same -- which is just the oldest > >> version it is known to exist in (which, if we would ensure ALL > >> ancestry is present, then it will truly be the origin). Happy > >> birthday Tim Rowledge. > >> > >> I have done VERY minimal testing. But we need to finally begin the > >> process of moving this crucial component our community away from the > >> costly Hetzner box and onto Gandi-based servers provided by the SFC, > >> so please try it and provide feedback! > >> > >> To use the new prototype: > >> > >> 1) Use a recent trunk image. > >> 2) Install Monticello-cmm.565 from the Inbox. 'browse mc versions' > >> and 'browse mc origin' will be added to methods and class-list menus > >> (but are initially grayed out). > >> 3) Add this repository to Monticello: > >> > >> MCHttpRepository > >> location: 'http://box4.squeak.org:8888/trunk' > >> user: 'id' > >> password: 'pw' > >> > >> 4) Add the above repository to whichever packages you wish to invoke > >> the new SCM features. > >> > >> - Chris > >> > >> PS -- This prototype only has data loaded for the trunk project as of > >> about one week ago. It is not kept in-sync with production trunk, > >> this is just a test. > > |
> With Monticello-cmm.569 it seems to be doing the right things. Very nice!
> > One question - Browse mc versions gives a patch browser that displays the > versions, but it does not seem to be showing method timestamp and author > initial information, You have #annotationPanes preference enabled, don't you? If so, which method is it not showing up for? It seems to be working for me.. > or the mc origin of each method version. Is this > not yet implemented, or am I missing something in my image that is present > in yours? Do you mean a context-menu entry 'Browse mc origin' on the Patch Browser's context menu? I had not thought of that but yes, since each displayed PatchOperation knows its #definition, I suppose that could be done. It's not implemented but a work-around is simply select "install" on the one you want to know the origin, then select "Browse mc origin" from the Hierarchy or PackagePane browser. Don't forget to revert the method if you don't actually want it part of your image. |
On Sat, Sep 07, 2013 at 05:12:48PM -0500, Chris Muller wrote:
> > With Monticello-cmm.569 it seems to be doing the right things. Very nice! > > > > One question - Browse mc versions gives a patch browser that displays the > > versions, but it does not seem to be showing method timestamp and author > > initial information, > > You have #annotationPanes preference enabled, don't you? If so, which > method is it not showing up for? It seems to be working for me.. I see now. I was expecting the MCPatchBrowser to show the information in the same way as a VersionsBrowser, but it does not. Instead, the version labels are in the middle panel. My mistake. > > > or the mc origin of each method version. Is this > > not yet implemented, or am I missing something in my image that is present > > in yours? > > Do you mean a context-menu entry 'Browse mc origin' on the Patch > Browser's context menu? I had not thought of that but yes, since each > displayed PatchOperation knows its #definition, I suppose that could > be done. Yes, something like that :) > > It's not implemented but a work-around is simply select "install" on > the one you want to know the origin, then select "Browse mc origin" > from the Hierarchy or PackagePane browser. Don't forget to revert the > method if you don't actually want it part of your image. Ok, I see. Thanks for doing this, it's great! Dave |
Free forum by Nabble | Edit this page |