Hi Juan!
if anyone expects that he clearly never coded :))))))))))))))) You did a hell of a job. To my naive eyes what is needed is not a volunteer for a suicide mission, but rather some form of integration among the many (too many?) images. The advantage VW clearly has is in having ONE stable base with many options, here we have many a base, all with huge merits and none with all the basic answers a simple developer needs. In some way there is a strange "all by myself" approach, which is maybe dictated by the fact that all of us have our own image, work on it until we get really tired and eventually end up almost in insulation. In such a situation we all address what is relevant "for us", we all do the best we can but there seem to be little chances to integrate all this work into a common stable distro. Get me right, I'm not criticising people. I'm saying that the load of work that is being done doesn't eventually add up to something that a first time user perceives as "quality". This is something about the method, not about the people. Yet in the end this thing pushes away potential coders that could help in making things better. I fell in love with Smalltalk back in the 90s, when there was but VisualWorks around. I had loads of things I didn't like with VW, but I never had to spend months checking the same 10 lines on the like of 30 different images to see "which one makes a better luck". I have to do it with Squeak, and I do it because I do want to write Smalltalk code again and I'm ready to stand some pain in order to do it. If I had never written code in it I'd have probably concluded that this is but a waste of time. I'm not worried by the fact that I will have to stay up some more nights, because I perfectly know that I'll do it anyway. In instead, I'm worried by the fact that many people will get a wrong impression by their first contact. And I'm sure this very worry was YOUR worry when you started CUIS. CUIS is a wonderful distro, but today I had to inline a string of credits for a guy who has an Irish name, and in order to do it I'll have to make a chinese dance, because I cannot cut and paste his name into a string. Now go and explain to a paying customer that you are two weeks late because you cannot cut and paste a name... When Igor said this is just a toy I could only agree. Because this is what people who try to sell Squeak are condemned to hear over and over again, as long as these basic problems do not get solved. Once again, this does not mean that someone is lazy or stupid or an "enemy of the people". It means that *maybe* there could be a way to share the load of work you all do in a way that makes a better result with less effort. I underline it, I said *maybe* and I'm not able to suggest a remedy here and now. But surely talking about it in explicit terms can help to find such a remedy, if any exists. Berto 2009/4/16 Juan Vuletich <[hidden email]> Bert Freudenberg wrote: -- ============================== Constitution du 24 juin 1793 - Article 35. - Quand le gouvernement viole les droits du peuple, l'insurrection est, pour le peuple et pour chaque portion du peuple, le plus sacré des droits et le plus indispensable des devoirs. |
In reply to this post by Juan Vuletich-4
On 16.04.2009, at 12:26, Juan Vuletich wrote:
> This is an open source project. This is the developers community > ("squeak-dev", right?). You are not expected to complain and wait. > You are expected to help. True that, can't be said often enough. OTOH, for an open-source project with a history as long as ours it sounds pretty immature. You'ld think by now we could effort having users who just expect Squeak to work for developing their own stuff, rather than having to fix Squeak itself. - Bert - |
In reply to this post by Bèrto ëd Sèra
Hi Folks
Am 16.04.2009 um 13:38 schrieb Bèrto ëd Sèra: > CUIS is a wonderful distro, but today I had to inline a string of > credits for a guy who has an Irish name, and in order to do it I'll > have to make a chinese dance, because I cannot cut and paste his > name into a string. Now go and explain to a paying customer that you > are two weeks late because you cannot cut and paste a name... When > Igor said this is just a toy I could only agree. Because this is > what people who try to sell Squeak are condemned to hear over and > over again, as long as these basic problems do not get solved. > > Once again, this does not mean that someone is lazy or stupid or an > "enemy of the people". It means that *maybe* there could be a way to > share the load of work you all do in a way that makes a better > result with less effort. I underline it, I said *maybe* and I'm not > able to suggest a remedy here and now. But surely talking about it > in explicit terms can help to find such a remedy, if any exists. +1 Actually this "chinese dance" also caused me to stay away from Cuis, which I really liked in general. Especially the speed. Please do consider to have unicode as not optional. Side remark: I tried Pharo as well for reading in some data with unicode but it also coughed on Umlauts -- so I found myself to use the squeakland distro again, which for me actually works best -- as soon as I have changed it from kids-mode to developer-mode. Oh well... Cheers Markus |
In reply to this post by Sophie424
On Wed, Apr 15, 2009 at 9:11 PM, Sophie (itsme213) <[hidden email]> wrote:
> Is there any chance the Pharo folks would be open to exploring whether Cuis > can be used as a base for their work? That will never happen because Cuis derived from Squeak 3.7 IIRC. Pharo would have to first update things to 3.8 and 3.9 and then reapply all Pharo changes. -- Damien Cassou http://damiencassou.seasidehosting.st |
In reply to this post by Bert Freudenberg
Bert Freudenberg wrote:
> On 16.04.2009, at 12:26, Juan Vuletich wrote: > >> This is an open source project. This is the developers community >> ("squeak-dev", right?). You are not expected to complain and wait. >> You are expected to help. > > > True that, can't be said often enough. > > OTOH, for an open-source project with a history as long as ours it > sounds pretty immature. You'ld think by now we could effort having > users who just expect Squeak to work for developing their own stuff, > rather than having to fix Squeak itself. > > - Bert - > Cheers, Juan Vuletich |
In reply to this post by Markus Gälli-3
Markus Gaelli wrote:
> Hi Folks > > Am 16.04.2009 um 13:38 schrieb Bèrto ëd Sèra: > >> CUIS is a wonderful distro, but today I had to inline a string of >> credits for a guy who has an Irish name, and in order to do it I'll >> have to make a chinese dance, because I cannot cut and paste his name >> into a string. Now go and explain to a paying customer that you are >> two weeks late because you cannot cut and paste a name... When Igor >> said this is just a toy I could only agree. Because this is what >> people who try to sell Squeak are condemned to hear over and over >> again, as long as these basic problems do not get solved. >> >> Once again, this does not mean that someone is lazy or stupid or an >> "enemy of the people". It means that *maybe* there could be a way to >> share the load of work you all do in a way that makes a better result >> with less effort. I underline it, I said *maybe* and I'm not able to >> suggest a remedy here and now. But surely talking about it in >> explicit terms can help to find such a remedy, if any exists. > > +1 > > Actually this "chinese dance" also caused me to stay away from Cuis, > which I really liked in general. Especially the speed. > Please do consider to have unicode as not optional. > > Side remark: I tried Pharo as well for reading in some data with > unicode but it also coughed on Umlauts -- so I found myself to use the > squeakland distro again, which for me actually works best -- as soon > as I have changed it from kids-mode to developer-mode. Oh well... > > Cheers > > Markus "I hope all this works as an argument for considering Unicode optional in Cuis. Or does anybody expect that besides doing 5 years of cleaning towards a simpler and faster image, I also need to address all these issues myself? The whole idea of optional packages is to draw a boundary between the kernel and each of them, to allow different people to focus on different problems. " Cheers, Juan Vuletich |
In reply to this post by Bèrto ëd Sèra
Bèrto ëd Sèra wrote:
> Hi Juan! > ... (big snip) > > CUIS is a wonderful distro, but today I had to inline a string of > credits for a guy who has an Irish name, and in order to do it I'll > have to make a chinese dance, because I cannot cut and paste his name > into a string. Now go and explain to a paying customer that you are > two weeks late because you cannot cut and paste a name... When Igor > said this is just a toy I could only agree. Because this is what > people who try to sell Squeak are condemned to hear over and over > again, as long as these basic problems do not get solved. > > Once again, this does not mean that someone is lazy or stupid or an > "enemy of the people". It means that *maybe* there could be a way to > share the load of work you all do in a way that makes a better result > with less effort. I underline it, I said *maybe* and I'm not able to > suggest a remedy here and now. But surely talking about it in explicit > terms can help to find such a remedy, if any exists. > > Berto So it seems there are problems using Unicode on all Squeak distributions. What is needed is a group of people who care enough about the problem to fix it. I'm sure that code could be shared by various distributions. But somebody needs to do it. So, please, people worried about Unicode. Stand up, and volunteer. Make a team. Understand and fix the problem. Cheers, Juan Vuletich |
Hi Juan!
I slept like 2 hours in the last 48... and it won't get much better in the next 3 months. If nobody shows up in the meantime I can start to put some time on it in late summer, and if someone shows up before... PLEASE! Do write at least a blog about what you are doing. Berto 2009/4/16 Juan Vuletich <[hidden email]> Bčrto ėd Sčra wrote: -- ============================== Constitution du 24 juin 1793 - Article 35. - Quand le gouvernement viole les droits du peuple, l'insurrection est, pour le peuple et pour chaque portion du peuple, le plus sacré des droits et le plus indispensable des devoirs. |
In reply to this post by Bèrto ëd Sèra
Be`rto e"d Se`ra wrote:
> LOL Andreas... 99,999% of first-time users will NEVER fill a bug. They > will go out and say "It's pure bullshit". :))))) Come on... we all do it > ten times a day. Sure we do. But since you've been hanging around for a while I'm holding you to a higher standard ;-) > As per myself... I discussed this thing on IRC a number of times, wrote > about it here because under Win even directory paths prevent Squeak from > working and all I got was the usual kind of thing you get from unstable > Open Source stuff: not much really, but at least it didn't cost me a penny. Many people are not on IRC (I for sure am not). But to the best of my knowledge, all of these problems have been addressed. Your mail is a perfect example for the problems though; since I don't have a russian windows system I can't really test whether anything is broken! That's why reporting these issues, even if briefly ("I just tried your latest VM and it still seems to have problem x, y, or z") is so important. > No probs, I switched OS and images until I found a configuration that > works for what I need, but while I can use Squeak for development I > surely don't plan to distribute it. It is self-evident that the > community is currently too small to make a stable distro. I admire you > all for what you are doing, and I will contribute as much as I can, but > customers want things that do not break. For sure. But friends help and let you know what's broken ;-) Cheers, - Andreas |
In reply to this post by Bert Freudenberg
Bert Freudenberg wrote:
> Err, we know quite well about these problems. Typing unicode chars don't > work because the utf32 charcode is not used by the released image. To > paste unicode text on the Mac you would need the image support for the > extendend clipboard plugin. But even if typing or pasting works, the > fonts only cover the latin-1 range, not full unicode. Neither Freetype > nor HostFonts nor Pango support is in the released image, nor did we > standardize on one. > > All of these issues have been solved already, but did not make it into a > release afaik. Interesting. Goes to show that I haven't been paying much attention. But if the fixes are there (can you send some pointers?) it sounds like a perfect candidate to assemble those into 3.11 and get this stuff straight. For fonts, there is Dejavu (http://dejavu-fonts.org/) which is based on the Bitstream Vera family and should make for a virtually identical replacement. Cheers, - Andreas |
In reply to this post by Andreas.Raab
Hi!
Many people are not on IRC (I for sure am not). But to the best of my knowledge, all of these problems have been addressed. Your mail is a perfect example for the problems though; since I don't have a russian windows system I can't really test whether anything is broken! That's why reporting these issues, even if briefly ("I just tried your latest VM and it still seems to have problem x, y, or z") is so important. I should think (it actually reads "I do hope") that a localised path with umlauts could break it much in the same way, at least, what Markus wrote makes me hope it should. It's definitely a problem to verify each possible combination, and I'm so far not even venturing into imagining how you can make an Hebrew or Arabic interface for people who may need RTL rendered GUIs. I have a mixed Russian/English system now on Ubuntu and I still have my wife's vista laptop in Russian, so whenever you guys have a candidate release I can verify it in some 10 minutes. Sadly I won't have much spare time until late summer, but I can always find 10 minutes to make a simple test. For sure. But friends help and let you know what's broken ;-) Deal :) Berto |
In reply to this post by Bert Freudenberg
+1
Ken On Thu, 2009-04-16 at 13:38 +0200, Bert Freudenberg wrote: > On 16.04.2009, at 12:26, Juan Vuletich wrote: > > > This is an open source project. This is the developers community > > ("squeak-dev", right?). You are not expected to complain and wait. > > You are expected to help. > > > True that, can't be said often enough. > > OTOH, for an open-source project with a history as long as ours it > sounds pretty immature. You'ld think by now we could effort having > users who just expect Squeak to work for developing their own stuff, > rather than having to fix Squeak itself. > > - Bert - > > > signature.asc (196 bytes) Download Attachment |
In reply to this post by Sophie424
isn't it there a way to try it automatically? it is Smalltalk... I'm not saying it is going to work 100% but...
Maybe something like this: 1) Load each method of cuis that is different or does not exist in pharo/squeak 2) For each method that is in pharo/squeak that is not is cuis and it does not have reference, remove it 3) ssomething similar for variables, classes, etc 4) Try pharo/squeak and see how it works! :-)
Just an idea... Hernan. On Wed, Apr 15, 2009 at 4:11 PM, Sophie (itsme213) <[hidden email]> wrote: Is there any chance the Pharo folks would be open to exploring whether Cuis |
In reply to this post by Andreas.Raab
BTW, the situation doesn't seem to be quite as dire as one might assume.
Attached are two screenshots which were created by: a) Installing Dejavu first (download, click, choose "install font") b) Going to ru.wikipedia.org and el.wikipedia.org c) Copying and pasting the first article from the front page That is about the extent of functionality that I can verify and it does seem to work allright except from the Character cr glyph displayed in Dejavu (I'm not sure where that comes from). Perhaps half of the problems could be solved simply by installing Dejavu by default? Cheers, - Andreas Andreas Raab wrote: > Bert Freudenberg wrote: >> Err, we know quite well about these problems. Typing unicode chars >> don't work because the utf32 charcode is not used by the released >> image. To paste unicode text on the Mac you would need the image >> support for the extendend clipboard plugin. But even if typing or >> pasting works, the fonts only cover the latin-1 range, not full >> unicode. Neither Freetype nor HostFonts nor Pango support is in the >> released image, nor did we standardize on one. >> >> All of these issues have been solved already, but did not make it into >> a release afaik. > > Interesting. Goes to show that I haven't been paying much attention. But > if the fixes are there (can you send some pointers?) it sounds like a > perfect candidate to assemble those into 3.11 and get this stuff > straight. For fonts, there is Dejavu (http://dejavu-fonts.org/) which is > based on the Bitstream Vera family and should make for a virtually > identical replacement. > > Cheers, > - Andreas > > |
In reply to this post by Juan Vuletich-4
Hi Juan,
Juan Vuletich pravi: > Markus Gaelli wrote: >> Actually this "chinese dance" also caused me to stay away from Cuis, >> which I really liked in general. Especially the speed. >> Please do consider to have unicode as not optional. >> >> Side remark: I tried Pharo as well for reading in some data with >> unicode but it also coughed on Umlauts -- so I found myself to use the >> squeakland distro again, which for me actually works best -- as soon >> as I have changed it from kids-mode to developer-mode. Oh well... > "I hope all this works as an argument for considering Unicode optional > in Cuis. Or does anybody expect that besides doing 5 years of cleaning > towards a simpler and faster image, I also need to address all these > issues myself? > > The whole idea of optional packages is to draw a boundary between the > kernel and each of them, to allow different people to focus on different > problems. " I'm not quite sure that you can make Unicode support optional, that is, loadable as a separate package. Unicode is roughly made of two things: - internal Unicode support (Unicode characters, strings, encoding) - input (keyboard) and output (UI, fonts) While you can make internal support loadable, input/output relies on operating system and therefore on Squeak VM, which is harder to make loadable, if at all. But Pharo guys are doing quite an effort lately on input/output part of Unicode support, maybe you can join forces? Best regards Janko -- Janko Mivšek AIDA/Web Smalltalk Web Application Server http://www.aidaweb.si |
In reply to this post by Andreas.Raab
On 17.04.2009, at 06:41, Andreas Raab wrote:
> BTW, the situation doesn't seem to be quite as dire as one might > assume. Attached are two screenshots which were created by: > a) Installing Dejavu first (download, click, choose "install font") > b) Going to ru.wikipedia.org and el.wikipedia.org > c) Copying and pasting the first article from the front page > That is about the extent of functionality that I can verify and it > does seem to work allright except from the Character cr glyph > displayed in Dejavu (I'm not sure where that comes from). > > Perhaps half of the problems could be solved simply by installing > Dejavu by default? > > Cheers, > - Andreas Windows VM does paste unicode. On the Mac I get this: The first line shows DejaVuSans does work, has cyrillic glyphs. (DejaVuSans-Bold.ttf produces an error on import) The second line shows typing does not work, I typed a cyrillic Б but it is treated as 0 (the utf32 problem). Below that the paste from ru.wikipedia. These are actual question marks (ASCII 63), the Mac VM does not paste Unicode even if setting SqueakEncodingType to UTF-8 (only the extended clipboard plugin does this). This is using the latest release of course (Squeak3.10.2-7179mac.zip from squeak.org's front page). - Bert - > > Andreas Raab wrote: >> Bert Freudenberg wrote: >>> Err, we know quite well about these problems. Typing unicode chars >>> don't work because the utf32 charcode is not used by the released >>> image. To paste unicode text on the Mac you would need the image >>> support for the extendend clipboard plugin. But even if typing or >>> pasting works, the fonts only cover the latin-1 range, not full >>> unicode. Neither Freetype nor HostFonts nor Pango support is in >>> the released image, nor did we standardize on one. >>> >>> All of these issues have been solved already, but did not make it >>> into a release afaik. >> Interesting. Goes to show that I haven't been paying much >> attention. But if the fixes are there (can you send some pointers?) >> it sounds like a perfect candidate to assemble those into 3.11 and >> get this stuff straight. For fonts, there is Dejavu (http://dejavu-fonts.org/ >> ) which is based on the Bitstream Vera family and should make for a >> virtually identical replacement. >> Cheers, >> - Andreas > > <ru.wikipedia.png><el.wikipedia.png> Workspace.png (13K) Download Attachment |
Bert Freudenberg wrote:
> One of the problems, yes. But note your example only works because the > Windows VM does paste unicode. On the Mac I get this: And the plugin then return non-precomposed UTF8. There a couple of "interesting" things that need fixing for full unicode support, too many to quickly explain. Best would be to look at the differences in the Pharo packages. Michael |
In reply to this post by Andreas.Raab
Hi Andreas!
What's the image you are using? I had problems in Ubuntu as well, and it simply was Latin accented chars. If we state a candidate image for reference I can map the problems on it. Berto 2009/4/17 Andreas Raab <[hidden email]> BTW, the situation doesn't seem to be quite as dire as one might assume. Attached are two screenshots which were created by: -- ============================== Constitution du 24 juin 1793 - Article 35. - Quand le gouvernement viole les droits du peuple, l'insurrection est, pour le peuple et pour chaque portion du peuple, le plus sacré des droits et le plus indispensable des devoirs. |
In reply to this post by hernan.wilkinson
well... maybe something like this is better:
1) From a pharo/squeak image, for each class we want to merge (ie. the morph hierarchy) 1.a) for each method of that class 1.a.a) compare the method with the 3.7 version (asuming cuis started from 3.7) and cuis
1.a.b) if the are equal, continue 1.a.c) if they are different 1.a.c.a) if the one that changed is the pharo/squeak version, do nothing 1.a.c.b) if the one that changed is the cuis version, load it
1.a.c.c) if both changes, ask a human to merge them :-) 1.a.c) if the method is in pharo/squeak but not in cuis and does not have senders in pharo/squeak, remove the method 1.a.d) if the class is in pharo/squeak but not in cuis and does not have references, remove it
1.a.e) same thing with instance variables as with methods (this should be done before of merging methods) I did this once with VisualAge/Envy and worked pretty well... anyway, I'll see if I can write something just to see how many differences are...
Hernan On Thu, Apr 16, 2009 at 11:20 PM, Hernan Wilkinson <[hidden email]> wrote: isn't it there a way to try it automatically? it is Smalltalk... I'm not saying it is going to work 100% but... |
In reply to this post by Janko Mivšek
Hi Janko!
There is also another issue: complexity. Loadable things are often far to be straightforward as it may seem: you often end up in having mutually excluding loadable packages. OTOH a multi-byte string is less performing than a single-byte string and many users who live in the ASCII side of the planet may not be willing to pay the performance fee. But... this is SmallTalk, isn't it? Why can't we load the unicode thing everywhere and use polymorphism so that if I load ASCII data it performs as ASCII data? It sounds more like SmallTalk to me. And a thing that is everywhere is usually much better debugged. Berto |
Free forum by Nabble | Edit this page |