Hi all!
Just hacked up a load script for performing all registered removals on SM - 10 of them (don't forget Balloon3D - it isn't in the correct category).
Read more on http://map2.squeakfoundation.org/sm/packagebyname/aggregated
Tested by firing up a vanilla 3.6a-5179, "open package loader" and then install the above package. Answer yes twice and then you have an image somewhat smaller.
NOTE: I haven't tried to bring stuff back in yet... :-) But so far so good.
regards, Göran
goran.hultgren@bluefish.se wrote:
Just hacked up a load script for performing all registered removals on SM - 10 of them (don't forget Balloon3D - it isn't in the correct category).
Discussions with Andreas have revealed that there is at least one more class needing removing in the VMcode removal - B3DPoolDefiner - unless perchance it gets taken by one of the others anyway.
NOTE: I haven't tried to bring stuff back in yet... :-) But so far so good.
Ah, well, trying to get the VM stuff back in at the moment might make you think you'd made an aggravated package removal :-(
tim
On Thu, Apr 10, 2003 at 03:37:11PM +0100, goran.hultgren@bluefish.se wrote:
Hi all!
Just hacked up a load script for performing all registered removals on SM - 10 of them (don't forget Balloon3D - it isn't in the correct category).
Read more on http://map2.squeakfoundation.org/sm/packagebyname/aggregated
Tested by firing up a vanilla 3.6a-5179, "open package loader" and then install the above package. Answer yes twice and then you have an image somewhat smaller.
NOTE: I haven't tried to bring stuff back in yet... :-) But so far so good.
Hi Goran,
Nice... we should really think about *adding* this stuff to the image. Not everything in one big lump, but incrementally.
I suggest to start with the Tests-Removal package: This a) works for me (hehe! ;-) b) comes with tests (trivially) and c) does not interfere with any other project (e.g. KCP, MCP).
It simply (IMHO of course) does NOT MAKE ANY sense to pile this stuff up until we get a big unmanageable pile of interdependend changesets!
Marcus
Hi!
Marcus Denker marcus@ira.uka.de wrote:
Hi Goran,
Nice... we should really think about *adding* this stuff to the image. Not everything in one big lump, but incrementally.=20
I agree. The idea I had was to eventually build some kind of "test server" (still on my drawing board) and in any case I thought these removals needed testing "in combination" and not just one by one. I assume there is a risk that they somehow interact so that in the end some weird combination simply doesn't work.
But if we create the "verification" packages (those that test that an installed package works) that you described earlier for these removals - then perhaps that will be enough. The "dreaded" combination-bugs will probably be found and dealt with during alpha.
I am all for getting these into the update stream ASAP - I mean, we are in alpha land know - let's go! :-) Unfortunately I have been a bit occupied lately but i am getting more time Very Soon Now (we have a release tomorrow of a Dolphin product).
I suggest to start with the Tests-Removal package: This a) works for me (hehe! ;-) b) comes with tests (trivially) and c) does not interfere with any other project (e.g. KCP, MCP).
Why not?
It simply (IMHO of course) does NOT MAKE ANY sense to pile this stuff up until we get a big unmanageable pile of interdependend changesets!
That was not the intention with the aggregated script - it was meant as a first step to test these removals in combination more easily before applying them. But hey, getting them in is more important.
Why don't we just sit down (I could do it in a few days - or someone else like you today :-) ) and just load the aggregated script and then load them back one by one in at least *one* ordering. They have a good chance of working I guess and if they do we could smack them in there all of them and then create those verification packages you described.
If problems later arise we just fix them - *again* we are in alpha land and we have the stream (that Andreas wants us to use goddarnit :-)).
And if they don't work together we can back down and just add your Tests-Removal at least. Sounds good?
Doug? Daniel?
regards, Göran
On Tue, Apr 29, 2003 at 11:31:12AM +0100, goran.hultgren@bluefish.se wrote:
I agree. The idea I had was to eventually build some kind of "test server" (still on my drawing board) and in any case I thought these removals needed testing "in combination" and not just one by one. I assume there is a risk that they somehow interact so that in the end some weird combination simply doesn't work.
Yes, could be. But this is true for all packages on SqueakMap, not only the new ones. The only real solution would be something like classboxes... (calls additions can be made private your own package with classboxes).
But if we create the "verification" packages (those that test that an installed package works) that you described earlier for these removals - then perhaps that will be enough. The "dreaded" combination-bugs will probably be found and dealt with during alpha.
Ok, we need simple test packages: There is one for the BaseImage Tests removal. Why wait?
I suggest to start with the Tests-Removal package: This a) works for me (hehe! ;-) b) comes with tests (trivially) and c) does not interfere with any other project (e.g. KCP, MCP).
Why not?
Because this only removes SUnit-tests and some hand-written testing methods that have been rewritten as SUnit tests. If KCP/MCP provide tests, they will not go into the Image, but the BaseImage test packe instead...
If problems later arise we just fix them - *again* we are in alpha land and we have the stream (that Andreas wants us to use goddarnit :-)).
And if they don't work together we can back down and just add your Tests-Removal at least. Sounds good?
I would like to look at them one-by-one. Add one, test it. Then look at the next one...
Marcus
Hi Marcus!
Marcus Denker marcus@ira.uka.de wrote:
On Tue, Apr 29, 2003 at 11:31:12AM +0100, goran.hultgren@bluefish.se wrote:
I agree. The idea I had was to eventually build some kind of "test server" (still on my drawing board) and in any case I thought these removals needed testing "in combination" and not just one by one. I assume there is a risk that they somehow interact so that in the end some weird combination simply doesn't work.
Yes, could be. But this is true for all packages on SqueakMap, not only the new ones.
True. But the difference is that the net effect is that we break stuff. :-)
The only real solution would be something like classboxes... (calls additions can be made private your own package with classboxes).=20
But if we create the "verification" packages (those that test that an installed package works) that you described earlier for these removals - then perhaps that will be enough. The "dreaded" combination-bugs will probably be found and dealt with during alpha.
Ok, we need simple test packages: There is one for the BaseImage Tests removal. Why wait?
I am not advocating to wait (I think you misunderstood my last post)!
I suggest to start with the Tests-Removal package: This a) works for me (hehe! ;-) b) comes with tests (trivially) and c) does not interfere with any other project (e.g. KCP, MCP).
Why not?
(this is probably where we misunderstood each other - my "Why not?" was meant as "Sure, why not - let's go ahead!" - and not meant as... well, I don't know - did you miss the "not" word? :-))
Because this only removes SUnit-tests and some hand-written testing methods that have been rewritten as SUnit tests. If KCP/MCP provide tests, they will not go into the Image, but the BaseImage test packe instead...
If problems later arise we just fix them - *again* we are in alpha land and we have the stream (that Andreas wants us to use goddarnit :-)).=20 =20 And if they don't work together we can back down and just add your Tests-Removal at least. Sounds good? =20
I would like to look at them one-by-one. Add one, test it. Then look at the next one...=20
Fine by me. Daniel also wrote that. So, who does it? It's all about harvesting.
Unfortunately I don't have the time right now and even if I had I should be focusing on SM1.1 anyway.
Marcus - if you have the time, why don't you simply sign up as a harvester and help out doing it? Perhaps you are already signed up. ;-)
I think all others agree with me on this (Doug?) - let's just get on with the removals one by one.
regards, Göran
On Wed, Apr 30, 2003 at 10:40:03AM +0100, goran.hultgren@bluefish.se wrote:
Why not?
(this is probably where we misunderstood each other - my "Why not?" was meant as "Sure, why not - let's go ahead!" - and not meant as... well, I don't know - did you miss the "not" word? :-))
;-) just a parse error.
Fine by me. Daniel also wrote that. So, who does it? It's all about harvesting.
Unfortunately I don't have the time right now and even if I had I should be focusing on SM1.1 anyway.
Marcus - if you have the time, why don't you simply sign up as a harvester and help out doing it? Perhaps you are already signed up. ;-)
Where do I need to sign up?
Hi Marcus!
Marcus Denker marcus@ira.uka.de wrote:
On Wed, Apr 30, 2003 at 10:40:03AM +0100, goran.hultgren@bluefish.se wrote:
Why not?
(this is probably where we misunderstood each other - my "Why not?" was meant as "Sure, why not - let's go ahead!" - and not meant as... well, I don't know - did you miss the "not" word? :-))
;-) just a parse error.=20
Okidoki.
[SNIP]
Marcus - if you have the time, why don't you simply sign up as a harvester and help out doing it? Perhaps you are already signed up. ;-)
Where do I need to sign up?
Well, take a look at: http://swiki.squeakfoundation.org/squeakfoundation/99
And related pages. Essentially you just post on the sqf-list and Doug will pick you up.
IMHO we simply need more Harvesters - any whoever is competent (and you definitely are) and willing should (again IMHO) be very welcome to help out. Harvesting was never meant to be an activity of just the Guides nor IMHO just a selected few. As long as we keep the competence level appropriately high I can't see any problem with more Harvesters.
regards, Göran
Hi Göran,
Well, take a look at: http://swiki.squeakfoundation.org/squeakfoundation/99
And related pages. Essentially you just post on the sqf-list and Doug will pick you up.
IMHO we simply need more Harvesters - any whoever is competent (and you definitely are) and willing should (again IMHO) be very welcome to help out.
What about to make a "oficial" call to help?
I think a lot of us are able to work in this area and I never know the "door is open".
Harvesting was never meant to be an activity of just the Guides nor IMHO just a selected few. As long as we keep the competence level appropriately high I can't see any problem with more Harvesters.
regards, Göran
Cheers,
Diego Gomez Deck
On Wednesday, April 30, 2003, at 05:40 AM, goran.hultgren@bluefish.se wrote:
I suggest to start with the Tests-Removal package: This a) works for me (hehe! ;-) b) comes with tests (trivially) and c) does not interfere with any other project (e.g. KCP, MCP).
Why not?
(this is probably where we misunderstood each other - my "Why not?" was meant as "Sure, why not - let's go ahead!" - and not meant as... well, I don't know - did you miss the "not" word? :-))
No big deal, but I interpreted the "Why not" in the same literal way Marcus did. :-) As in, "Why does the Tests-Removal package not interfere with any other project?" If you meant it as in the more rhetorical "Sure, why not", be sure to include the "Sure" next time. This concludes our English colloquialism lesson for the day. ;-)
- Doug Way
squeak-dev@lists.squeakfoundation.org