[ANN] Squeak3.9g-7054...
Locked
 
|
Hi,
Ok - FFI support is removed. But how to get it back? I havent found the Monticello packages on the usual places (PackageLoader/SqueakMap, SqueakSource.com) I was lucky to see the mcz still be available from http://source.squeakfoundation.org/39a/ but it was not loadable in the first place. If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is required. Any newbee loading one of these packages using PackageLoader will run into trouble, declare Squeak as useless and will soon delete Squeak from it's harddisk! This will definitely not help the community to grow - instead we may end up with only a few people (most of them burned out like Marcus) :( I feel a little bit sad that there is no reliable process behind all that - next time we may decide to remove Morphic in the last minute. There may be reasons for Andreas to get FFI removed (IMHO I agree that FFI should not be a part of the base image) but why speak up so late? Dont know how usefull the initial "Release Plan for 3.9" really was ... (see http://minnow.cc.gatech.edu/squeak/5645) However - I'm happy that a new version is out. :) Thanks for all the hard work! bye Torsten -- Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer |
Re: [ANN] Squeak3.9g-7054...
|
|
Let us Andreas propose a solution because he is the main packager of
FFI. Else I will roll back to 7053 and ship 3.9 with FFI > Ok - FFI support is removed. But how to get it back? I havent found > the Monticello packages on the usual places (PackageLoader/SqueakMap, > SqueakSource.com) > > I was lucky to see the mcz still be available from > http://source.squeakfoundation.org/39a/ but it was not loadable > in the first place. What was the problem? > If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is > required. > > Any newbee loading one of these packages using PackageLoader will > run into > trouble, declare Squeak as useless and will soon delete Squeak from > it's harddisk! > This will definitely not help the community to grow - instead we > may end > up with only a few people (most of them burned out like Marcus) :( agreed > I feel a little bit sad that there is no reliable process behind > all that - > next time we may decide to remove Morphic in the last minute. We tried. But this is difficult when the communication between people is not easy. > There may be reasons for Andreas to get FFI removed (IMHO I agree > that FFI > should not be a part of the base image) but why speak up so late? > > Dont know how usefull the initial "Release Plan for 3.9" really > was ... > (see http://minnow.cc.gatech.edu/squeak/5645) At least it was for us as a goal and we are pretty there. No? > However - I'm happy that a new version is out. :) > Thanks for all the hard work! > > bye > Torsten > > -- > > > Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! > Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer > |
|
|
In reply to this post by Torsten Bergmann
Sorry - it'll come right up. I was away on the weekend and am still on
my way home (writing this from Starbucks at SEA waiting form my delayed flight home). Cheers, - Andreas [hidden email] wrote: > Hi, > > Ok - FFI support is removed. But how to get it back? I havent found > the Monticello packages on the usual places (PackageLoader/SqueakMap, > SqueakSource.com) > > I was lucky to see the mcz still be available from > http://source.squeakfoundation.org/39a/ but it was not loadable > in the first place. > > If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is required. > > Any newbee loading one of these packages using PackageLoader will run into > trouble, declare Squeak as useless and will soon delete Squeak from it's harddisk! > This will definitely not help the community to grow - instead we may end > up with only a few people (most of them burned out like Marcus) :( > > I feel a little bit sad that there is no reliable process behind all that - > next time we may decide to remove Morphic in the last minute. > > There may be reasons for Andreas to get FFI removed (IMHO I agree that FFI > should not be a part of the base image) but why speak up so late? > > Dont know how usefull the initial "Release Plan for 3.9" really was ... > (see http://minnow.cc.gatech.edu/squeak/5645) > > However - I'm happy that a new version is out. :) > Thanks for all the hard work! > > bye > Torsten > |
FFI on Squeakmap (was: Re: [ANN] Squeak3.9g-7054...)
|
|
In reply to this post by Torsten Bergmann
Hi -
The FFI is now up at SqueakMap. Grab the latest 3.9 "update map from the net" and you should be able to load it without any problems. If you (or anyone else) encounters problems, please file a bug - I think I got the installation issues nailed but more testing is welcome. I have also used this update to make a few more changes like: * Restructuring the FFI packages: This was required since the kernel needs to be loaded first such that the rest (tests+examples) can be parsed properly. * Some "interesting" initialization fixes to make it possible to load the FFI-Kernel package properly. Obviously nobody has ever to load/unload those before - there were bugs in there that made even me blush ;-) * SUnit-ized the existing tests: While noting that the "FFI-Plugin" category contained only tests, I changed it to "FFI-Tests" and rewrote the existing tests to be SUnit style. There is one problem left, namely that an unload of the FFI will only be clean if and only if PackageInfo is updated first. 3.9 contains an outdated version of PI which does not support scripts properly and while loading seems to work okay (mostly because the load script will be executed even if not preserved) unloading will require manual cleanup. Can't be helped at this point. Cheers, - Andreas [hidden email] wrote: > Hi, > > Ok - FFI support is removed. But how to get it back? I havent found > the Monticello packages on the usual places (PackageLoader/SqueakMap, > SqueakSource.com) > > I was lucky to see the mcz still be available from > http://source.squeakfoundation.org/39a/ but it was not loadable > in the first place. > > If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is required. > > Any newbee loading one of these packages using PackageLoader will run into > trouble, declare Squeak as useless and will soon delete Squeak from it's harddisk! > This will definitely not help the community to grow - instead we may end > up with only a few people (most of them burned out like Marcus) :( > > I feel a little bit sad that there is no reliable process behind all that - > next time we may decide to remove Morphic in the last minute. > > There may be reasons for Andreas to get FFI removed (IMHO I agree that FFI > should not be a part of the base image) but why speak up so late? > > Dont know how usefull the initial "Release Plan for 3.9" really was ... > (see http://minnow.cc.gatech.edu/squeak/5645) > > However - I'm happy that a new version is out. :) > Thanks for all the hard work! > > bye > Torsten > |
Re: FFI on Squeakmap (was: Re: [ANN] Squeak3.9g-7054...)
|
|
On 15 août 06, at 09:01, Andreas Raab wrote: > Hi - > > The FFI is now up at SqueakMap. Grab the latest 3.9 "update map > from the net" and you should be able to load it without any > problems. If you (or anyone else) encounters problems, please file > a bug - I think I got the installation issues nailed but more > testing is welcome. > > I have also used this update to make a few more changes like: Cool. > > * Restructuring the FFI packages: This was required since the > kernel needs to be loaded first such that the rest (tests+examples) > can be parsed properly. > > * Some "interesting" initialization fixes to make it possible to > load the FFI-Kernel package properly. Obviously nobody has ever to > load/unload those before - there were bugs in there that made even > me blush ;-) :) I tried but I thought this was my ununderstanding of the problem. > * SUnit-ized the existing tests: While noting that the "FFI-Plugin" > category contained only tests, I changed it to "FFI-Tests" and > rewrote the existing tests to be SUnit style. > > There is one problem left, namely that an unload of the FFI will > only be clean if and only if PackageInfo is updated first. 3.9 > contains an outdated version of PI which does not support scripts > properly and while loading seems to work okay (mostly because the > load script will be executed even if not preserved) unloading will > require manual cleanup. Can't be helped at this point. Do you know how to do that? Goran which version of PackageInfo should I load. Stef > > Cheers, > - Andreas > > [hidden email] wrote: >> Hi, >> Ok - FFI support is removed. But how to get it back? I havent found >> the Monticello packages on the usual places (PackageLoader/ >> SqueakMap, SqueakSource.com) >> I was lucky to see the mcz still be available from http:// >> source.squeakfoundation.org/39a/ but it was not loadable in the >> first place. >> If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is >> required. Any newbee loading one of these packages using >> PackageLoader will run into trouble, declare Squeak as useless and >> will soon delete Squeak from it's harddisk! This will definitely >> not help the community to grow - instead we may end >> up with only a few people (most of them burned out like Marcus) :( >> I feel a little bit sad that there is no reliable process behind >> all that - next time we may decide to remove Morphic in the last >> minute. There may be reasons for Andreas to get FFI removed (IMHO >> I agree that FFI >> should not be a part of the base image) but why speak up so late? >> Dont know how usefull the initial "Release Plan for 3.9" really >> was ... >> (see http://minnow.cc.gatech.edu/squeak/5645) >> However - I'm happy that a new version is out. :) >> Thanks for all the hard work! bye >> Torsten > > |
PackageInfo in 3.9 (was: Re: FFI on Squeakmap)
|
|
stéphane ducasse wrote:
>> There is one problem left, namely that an unload of the FFI will only >> be clean if and only if PackageInfo is updated first. 3.9 contains an >> outdated version of PI which does not support scripts properly and >> while loading seems to work okay (mostly because the load script will >> be executed even if not preserved) unloading will require manual >> cleanup. Can't be helped at this point. > > Do you know how to do that? > Goran which version of PackageInfo should I load. I don't. One of the problems is that the packages in the image do not have their "home" location associated with it, and I don't know where the primary repository for PI is (compare this for example with the FFI where, after installing it from SM, Monticello points straight to source.sqf.org). This should be fixed - even though the 3.9a repository may be necessary for integration, packages should definitely have their "primary" repositories associated with them. Cheers, - Andreas |
Re: [ANN] Squeak3.9g-7054...
|
|
In reply to this post by stéphane ducasse-2
Hi
There's something very wrong with the sources or changes file. - take a Squeak3.9g-7054 image with SqueakV39.sources - open a brower on TraitDescription - display the method #addSelector:withMethod:notifying: -> oops decompiled code! There are several of these methods. You can find them by loading the NewCompiler and recompiling the image. The reported problemMethods are the ones with missing source. Philippe 2006/8/14, stéphane ducasse <[hidden email]>: > Let us Andreas propose a solution because he is the main packager of > FFI. > Else I will roll back to 7053 and ship 3.9 with FFI > > > Ok - FFI support is removed. But how to get it back? I havent found > > the Monticello packages on the usual places (PackageLoader/SqueakMap, > > SqueakSource.com) > > > > I was lucky to see the mcz still be available from > > http://source.squeakfoundation.org/39a/ but it was not loadable > > in the first place. > > What was the problem? > > > If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is > > required. > > > > Any newbee loading one of these packages using PackageLoader will > > run into > > trouble, declare Squeak as useless and will soon delete Squeak from > > it's harddisk! > > This will definitely not help the community to grow - instead we > > may end > > up with only a few people (most of them burned out like Marcus) :( > > agreed > > > I feel a little bit sad that there is no reliable process behind > > all that - > > next time we may decide to remove Morphic in the last minute. > > We tried. > But this is difficult when the communication between people is not easy. > > > There may be reasons for Andreas to get FFI removed (IMHO I agree > > that FFI > > should not be a part of the base image) but why speak up so late? > > > > Dont know how usefull the initial "Release Plan for 3.9" really > > was ... > > (see http://minnow.cc.gatech.edu/squeak/5645) > > At least it was for us as a goal and we are pretty there. No? > > > However - I'm happy that a new version is out. :) > > Thanks for all the hard work! > > > > bye > > Torsten > > > > -- > > > > > > Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! > > Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer > > > > > |
Re: [ANN] Squeak3.9g-7054...
|
|
Oops.
Thanks Stef > > There's something very wrong with the sources or changes file. > - take a Squeak3.9g-7054 image with SqueakV39.sources > - open a brower on TraitDescription > - display the method #addSelector:withMethod:notifying: > > -> oops decompiled code! > > There are several of these methods. You can find them by loading the > NewCompiler and recompiling the image. The reported problemMethods are > the ones with missing source. > > Philippe > > > 2006/8/14, stéphane ducasse <[hidden email]>: >> Let us Andreas propose a solution because he is the main packager of >> FFI. >> Else I will roll back to 7053 and ship 3.9 with FFI >> >> > Ok - FFI support is removed. But how to get it back? I havent found >> > the Monticello packages on the usual places (PackageLoader/ >> SqueakMap, >> > SqueakSource.com) >> > >> > I was lucky to see the mcz still be available from >> > http://source.squeakfoundation.org/39a/ but it was not loadable >> > in the first place. >> >> What was the problem? >> >> > If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is >> > required. >> > >> > Any newbee loading one of these packages using PackageLoader will >> > run into >> > trouble, declare Squeak as useless and will soon delete Squeak from >> > it's harddisk! >> > This will definitely not help the community to grow - instead we >> > may end >> > up with only a few people (most of them burned out like Marcus) :( >> >> agreed >> >> > I feel a little bit sad that there is no reliable process behind >> > all that - >> > next time we may decide to remove Morphic in the last minute. >> >> We tried. >> But this is difficult when the communication between people is not >> easy. >> >> > There may be reasons for Andreas to get FFI removed (IMHO I agree >> > that FFI >> > should not be a part of the base image) but why speak up so late? >> > >> > Dont know how usefull the initial "Release Plan for 3.9" really >> > was ... >> > (see http://minnow.cc.gatech.edu/squeak/5645) >> >> At least it was for us as a goal and we are pretty there. No? >> >> > However - I'm happy that a new version is out. :) >> > Thanks for all the hard work! >> > >> > bye >> > Torsten >> > >> > -- >> > >> > >> > Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! >> > Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer >> > >> >> >> > > |
Re: [ANN] Squeak3.9g-7054...
|
|
In fact more
"Arghhhhhhhhhhhhhhhhh" Something wrong in condenseSources with traits. Arghhhhhhhhhhhh Stef > Oops. > Thanks > > Stef > >> >> There's something very wrong with the sources or changes file. >> - take a Squeak3.9g-7054 image with SqueakV39.sources >> - open a brower on TraitDescription >> - display the method #addSelector:withMethod:notifying: >> >> -> oops decompiled code! >> >> There are several of these methods. You can find them by loading the >> NewCompiler and recompiling the image. The reported problemMethods >> are >> the ones with missing source. >> >> Philippe >> >> >> 2006/8/14, stéphane ducasse <[hidden email]>: >>> Let us Andreas propose a solution because he is the main packager of >>> FFI. >>> Else I will roll back to 7053 and ship 3.9 with FFI >>> >>> > Ok - FFI support is removed. But how to get it back? I havent >>> found >>> > the Monticello packages on the usual places (PackageLoader/ >>> SqueakMap, >>> > SqueakSource.com) >>> > >>> > I was lucky to see the mcz still be available from >>> > http://source.squeakfoundation.org/39a/ but it was not loadable >>> > in the first place. >>> >>> What was the problem? >>> >>> > If one wants to use SQLite, ODBC, ExternalWebBrowser, ... FFI is >>> > required. >>> > >>> > Any newbee loading one of these packages using PackageLoader will >>> > run into >>> > trouble, declare Squeak as useless and will soon delete Squeak >>> from >>> > it's harddisk! >>> > This will definitely not help the community to grow - instead we >>> > may end >>> > up with only a few people (most of them burned out like Marcus) :( >>> >>> agreed >>> >>> > I feel a little bit sad that there is no reliable process behind >>> > all that - >>> > next time we may decide to remove Morphic in the last minute. >>> >>> We tried. >>> But this is difficult when the communication between people is >>> not easy. >>> >>> > There may be reasons for Andreas to get FFI removed (IMHO I agree >>> > that FFI >>> > should not be a part of the base image) but why speak up so late? >>> > >>> > Dont know how usefull the initial "Release Plan for 3.9" really >>> > was ... >>> > (see http://minnow.cc.gatech.edu/squeak/5645) >>> >>> At least it was for us as a goal and we are pretty there. No? >>> >>> > However - I'm happy that a new version is out. :) >>> > Thanks for all the hard work! >>> > >>> > bye >>> > Torsten >>> > >>> > -- >>> > >>> > >>> > Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! >>> > Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer >>> > >>> >>> >>> >> >> > > > |
Re: [ANN] Squeak3.9g-7054...
|
|
stéphane ducasse puso en su mail :
> In fact more > "Arghhhhhhhhhhhhhhhhh" > > Something wrong in condenseSources with traits. Arghhhhhhhhhhhh > > Stef The Pavel fix for condenseSources work in 7045. (needed in the process of build the KernelSqueak.7045) I don't know if works in newer, tomorrow check in my in use .image and tell. Edgar __________________________________________________ Preguntá. Respondé. Descubrí. Todo lo que querías saber, y lo que ni imaginabas, está en Yahoo! Respuestas (Beta). ¡Probalo ya! http://www.yahoo.com.ar/respuestas |
Re: [ANN] Squeak3.9g-7054...
|
|
Hi edgar
which fix? Because adrian under at the military service for two weeks and I know that he fixes the condenseChanges so may be this is the same problem for condenseSources. Bad timing. Stef On 17 août 06, at 00:16, Lic. Edgar J. De Cleene wrote: > stéphane ducasse puso en su mail : > >> In fact more >> "Arghhhhhhhhhhhhhhhhh" >> >> Something wrong in condenseSources with traits. Arghhhhhhhhhhhh >> >> Stef > The Pavel fix for condenseSources work in 7045. (needed in the > process of > build the KernelSqueak.7045) > I don't know if works in newer, tomorrow check in my in use .image > and tell. > > > Edgar > > > > > > > __________________________________________________ > Preguntá. Respondé. Descubrí. > Todo lo que querías saber, y lo que ni imaginabas, > está en Yahoo! Respuestas (Beta). > ¡Probalo ya! > http://www.yahoo.com.ar/respuestas > > > |
|
|
Hi,
I have never created fix for condenseChanges or condenseSrouces. I used code from this Mantis record: http://bugs.impara.de/view.php?id=3737 -- Pavel > Hi edgar > > which fix? > Because adrian under at the military service for two weeks and I know > that he fixes the condenseChanges so may be this is the same > problem for condenseSources. Bad timing. > > Stef > > On 17 août 06, at 00:16, Lic. Edgar J. De Cleene wrote: > > > stéphane ducasse puso en su mail : > > > >> In fact more > >> "Arghhhhhhhhhhhhhhhhh" > >> > >> Something wrong in condenseSources with traits. Arghhhhhhhhhhhh > >> > >> Stef > > The Pavel fix for condenseSources work in 7045. (needed in the > > process of > > build the KernelSqueak.7045) > > I don't know if works in newer, tomorrow check in my in use .image > > and tell. > > > > > > Edgar > > > > > > > > > > > > > > __________________________________________________ > > Preguntá. Respondé. Descubrí. > > Todo lo que querías saber, y lo que ni imaginabas, > > está en Yahoo! Respuestas (Beta). > > ¡Probalo ya! > > http://www.yahoo.com.ar/respuestas > > > > > > > > > |
Re: [ANN] Squeak3.9g-7054...
|
|
In reply to this post by stéphane ducasse-2
stéphane ducasse puso en su mail :
> Hi edgar > > which fix? > Because adrian under at the military service for two weeks and I know > that he fixes the condenseChanges so may be this is the same > problem for condenseSources. Bad timing. > > Stef I in UTN, so I don't have to hand now. The short story is, as image builder person for long time, you always have troubles when you produce a new sources. I have this problem in SqueakLight experiments. So I learn to use "official" SqueakV3.sources and things improve. But sometimes this hand builded images could complaint with condenseChanges and shows same problem what was reported. Until Pavel need solve the problem or he do not could produce "our" kernel. And I using his fix on all my experiments and works. Also I do in 3.9 7045 (with traits) and works. I have now in use a 7051 full of things , so I do a backup copy, condenseSources and tell what happen. Wait six hours or search this archive list for Pavel fix and faster solution. Edgar __________________________________________________ Preguntá. Respondé. Descubrí. Todo lo que querías saber, y lo que ni imaginabas, está en Yahoo! Respuestas (Beta). ¡Probalo ya! http://www.yahoo.com.ar/respuestas |
Re: [ANN] Squeak3.9g-7054...
|
|
In reply to this post by Pavel Krivanek
Thanks pavel
Adrian used this fixes and certainly did more. So I will have to look at condenseSources as soon as I have a moment. Stef On 17 août 06, at 10:31, Pavel Krivanek wrote: > Hi, > > I have never created fix for condenseChanges or condenseSrouces. I > used code from this Mantis record: > http://bugs.impara.de/view.php?id=3737 > > -- Pavel > >> Hi edgar >> >> which fix? >> Because adrian under at the military service for two weeks and I know >> that he fixes the condenseChanges so may be this is the same >> problem for condenseSources. Bad timing. >> >> Stef >> >> On 17 août 06, at 00:16, Lic. Edgar J. De Cleene wrote: >> >> > stéphane ducasse puso en su mail : >> > >> >> In fact more >> >> "Arghhhhhhhhhhhhhhhhh" >> >> >> >> Something wrong in condenseSources with traits. Arghhhhhhhhhhhh >> >> >> >> Stef >> > The Pavel fix for condenseSources work in 7045. (needed in the >> > process of >> > build the KernelSqueak.7045) >> > I don't know if works in newer, tomorrow check in my in use .image >> > and tell. >> > >> > >> > Edgar >> > >> > >> > >> > >> > >> > >> > __________________________________________________ >> > Preguntá. Respondé. Descubrí. >> > Todo lo que querías saber, y lo que ni imaginabas, >> > está en Yahoo! Respuestas (Beta). >> > ¡Probalo ya! >> > http://www.yahoo.com.ar/respuestas >> > >> > >> > >> >> >> > > |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |