comments comments comments again
Locked
comments comments comments again
 
|
Hi
why can we put comment in class comment? For example I worked on fixing 1.2 and now I want to execute the configurationOfPharo and no magic invocation.... JUST A FUCKING EMPTY class comment as usual. So again I feel like an idiot and I'm trying in the dark. You have NOT A SINGLE REASON NOT TO PUT COMMENT! One of these days I will hack the system to break the load when a class as no comment. So now after randomly trying.....I get a nice DNU. Apparently my time is not worth anything.... So if you want to see if my fix works. Give a try I will not and if what I did break, may be I will have a look or may be not because no time for that after. And BTW do not put any comment, it feels much cooler that way. Stef |
Re: comments comments comments again
|
|
Since I have edited ConfigurationOfPharo I can add an appropriate comment ... I will add comments to the other configurations as well ...
Dale On Feb 27, 2011, at 7:18 AM, Stéphane Ducasse wrote: > Hi > > why can we put comment in class comment? > For example I worked on fixing 1.2 and now I want to execute the configurationOfPharo and no magic invocation.... > JUST A FUCKING EMPTY class comment as usual. > > > So again I feel like an idiot and I'm trying in the dark. You have NOT A SINGLE REASON NOT TO PUT COMMENT! > One of these days I will hack the system to break the load when a class as no comment. > > So now after randomly trying.....I get a nice DNU. > Apparently my time is not worth anything.... > > So if you want to see if my fix works. Give a try I will not and if what I did break, may be I will have a look or may be not > because no time for that after. > > > And BTW do not put any comment, it feels much cooler that way. > > Stef > > |
Re: comments comments comments again
|
|
Thanks dale.
Stef > Since I have edited ConfigurationOfPharo I can add an appropriate comment ... I will add comments to the other configurations as well ... > > Dale > > On Feb 27, 2011, at 7:18 AM, Stéphane Ducasse wrote: > >> Hi >> >> why can we put comment in class comment? >> For example I worked on fixing 1.2 and now I want to execute the configurationOfPharo and no magic invocation.... >> JUST A FUCKING EMPTY class comment as usual. >> >> >> So again I feel like an idiot and I'm trying in the dark. You have NOT A SINGLE REASON NOT TO PUT COMMENT! >> One of these days I will hack the system to break the load when a class as no comment. >> >> So now after randomly trying.....I get a nice DNU. >> Apparently my time is not worth anything.... >> >> So if you want to see if my fix works. Give a try I will not and if what I did break, may be I will have a look or may be not >> because no time for that after. >> >> >> And BTW do not put any comment, it feels much cooler that way. >> >> Stef >> >> > > |
Re: comments comments comments again
|
|
In reply to this post by Stéphane Ducasse
On Sun, Feb 27, 2011 at 4:18 PM, Stéphane Ducasse <[hidden email]> wrote: Hi I was the original author of such class. And yes, I don't write class comments. I am not use too, usually because: - I almost don't find it use - The browser doesn't help me to see class comments - Who can guess that under a '?' there are class comments? I discovered that YEARS after I was already using Smalltalk - I prefer writting/reading a TestCase for such class (I know it is not the same as class comments, but it helps). - I prefer looking at senders/references to such class - When class comments are long, the browser is FUCKING slow to write Even more, in this case, what would you write for example? ConfigurationOf* are just Metacello Configurations. What would you need to know about that despite of Metacello? If you know Metacello, you should know how to query the configuration and get all the things you want, like groups, packages, etc So now after randomly trying.....I get a nice DNU. But do you think that could be fixed just because of having class comments? Apparently my time is not worth anything.... |
|
|
I find it frustrating to load a package and to have no clue where to
start. The user must scan the classes and guess which one are 'entry points'. Class comments does not help there. For a user perspective I would prefer to have a HelpSystem indicating where to start. Now I guess class comments are useful when you're hacking/refactoring. On 27/02/11 19:49, Mariano Martinez Peck wrote: > > > On Sun, Feb 27, 2011 at 4:18 PM, Stéphane Ducasse > <[hidden email] <mailto:[hidden email]>> wrote: > > Hi > > why can we put comment in class comment? > For example I worked on fixing 1.2 and now I want to execute the > configurationOfPharo and no magic invocation.... > JUST A FUCKING EMPTY class comment as usual. > > > So again I feel like an idiot and I'm trying in the dark. You have > NOT A SINGLE REASON NOT TO PUT COMMENT! > One of these days I will hack the system to break the load when a > class as no comment. > > > I was the original author of such class. And yes, I don't write class > comments. I am not use too, usually because: > > - I almost don't find it use > - The browser doesn't help me to see class comments > - Who can guess that under a '?' there are class comments? I discovered > that YEARS after I was already using Smalltalk > - I prefer writting/reading a TestCase for such class (I know it is not > the same as class comments, but it helps). > - I prefer looking at senders/references to such class > - When class comments are long, the browser is FUCKING slow to write > > Even more, in this case, what would you write for example? > ConfigurationOf* are just Metacello Configurations. What would you need > to know about that despite of Metacello? If you know Metacello, you > should know how to query the configuration and get all the things you > want, like groups, packages, etc > > > So now after randomly trying.....I get a nice DNU. > > > But do you think that could be fixed just because of having class comments? > > Apparently my time is not worth anything.... > > So if you want to see if my fix works. Give a try I will not and if > what I did break, may be I will have a look or may be not > because no time for that after. > > > And BTW do not put any comment, it feels much cooler that way. > > Stef > > > |
Re: comments comments comments again
|
|
On Sun, Feb 27, 2011 at 10:03 PM, Francois Stephany <[hidden email]> wrote: I find it frustrating to load a package and to have no clue where to start. The user must scan the classes and guess which one are 'entry points'. Class comments does not help there. +99 Exactly. The few times I needed class comments, the problem was exactly that one: which classes, which where the entry point. Now I guess class comments are useful when you're hacking/refactoring. |
Re: comments comments comments again
|
|
In reply to this post by Mariano Martinez Peck
On Feb 27, 2011, at 7:49 PM, Mariano Martinez Peck wrote: > > > On Sun, Feb 27, 2011 at 4:18 PM, Stéphane Ducasse <[hidden email]> wrote: > Hi > > why can we put comment in class comment? > For example I worked on fixing 1.2 and now I want to execute the configurationOfPharo and no magic invocation.... > JUST A FUCKING EMPTY class comment as usual. > > > So again I feel like an idiot and I'm trying in the dark. You have NOT A SINGLE REASON NOT TO PUT COMMENT! > One of these days I will hack the system to break the load when a class as no comment. > > > I was the original author of such class. And yes, I don't write class comments. I am not use too, usually because: > > - I almost don't find it use > - The browser doesn't help me to see class comments > - Who can guess that under a '?' there are class comments? I discovered that YEARS after I was already using Smalltalk > - I prefer writting/reading a TestCase for such class (I know it is not the same as class comments, but it helps). > - I prefer looking at senders/references to such class > - When class comments are long, the browser is FUCKING slow to write none of the above is a justification. " the minimal statement that should be in the comment: To load me, execute ((Smalltalk globals at: #ConfigurationOfPharo) project version: #stable) load " I added it I do not want to have to remember that in this project I should not use latest, whatever. In ten years from now I want to doubleclick and it loads. > Even more, in this case, what would you write for example? ConfigurationOf* are just Metacello Configurations. What would you need to know about that despite of Metacello? If you know Metacello, you should know how to query the configuration and get all the things you want, like groups, packages, etc sorry but there are plenty of expressions that can make the life of the maintainer easier. Just think 2 min and you will come with - how to validate - how to load core I do not want to have our book chapter open to load something. > So now after randomly trying.....I get a nice DNU. > > But do you think that could be fixed just because of having class comments? Yes I do that in all my configuration because I forgot all the time and my brain is better occupied on real problems not stuff that can be fixed by adding one comment. > Apparently my time is not worth anything.... > > So if you want to see if my fix works. Give a try I will not and if what I did break, may be I will have a look or may be not > because no time for that after. > > > And BTW do not put any comment, it feels much cooler that way. > > Stef > > > |
Re: comments comments comments again
|
|
In reply to this post by fstephany
On Feb 27, 2011, at 10:03 PM, Francois Stephany wrote: > I find it frustrating to load a package and to have no clue where to start. The user must scan the classes and guess which one are 'entry points'. Class comments does not help there. Because there are not good. Look at the one of ConfigurationOfMana plain bad but you can always load the configuration or the system when I'm building the configuration. > For a user perspective I would prefer to have a HelpSystem indicating where to start. Now I guess class comments are useful when you're hacking/refactoring. Not necessary. > > On 27/02/11 19:49, Mariano Martinez Peck wrote: >> >> >> On Sun, Feb 27, 2011 at 4:18 PM, Stéphane Ducasse >> <[hidden email] <mailto:[hidden email]>> wrote: >> >> Hi >> >> why can we put comment in class comment? >> For example I worked on fixing 1.2 and now I want to execute the >> configurationOfPharo and no magic invocation.... >> JUST A FUCKING EMPTY class comment as usual. >> >> >> So again I feel like an idiot and I'm trying in the dark. You have >> NOT A SINGLE REASON NOT TO PUT COMMENT! >> One of these days I will hack the system to break the load when a >> class as no comment. >> >> >> I was the original author of such class. And yes, I don't write class >> comments. I am not use too, usually because: >> >> - I almost don't find it use >> - The browser doesn't help me to see class comments >> - Who can guess that under a '?' there are class comments? I discovered >> that YEARS after I was already using Smalltalk >> - I prefer writting/reading a TestCase for such class (I know it is not >> the same as class comments, but it helps). >> - I prefer looking at senders/references to such class >> - When class comments are long, the browser is FUCKING slow to write >> >> Even more, in this case, what would you write for example? >> ConfigurationOf* are just Metacello Configurations. What would you need >> to know about that despite of Metacello? If you know Metacello, you >> should know how to query the configuration and get all the things you >> want, like groups, packages, etc >> >> >> So now after randomly trying.....I get a nice DNU. >> >> >> But do you think that could be fixed just because of having class comments? >> >> Apparently my time is not worth anything.... >> >> So if you want to see if my fix works. Give a try I will not and if >> what I did break, may be I will have a look or may be not >> because no time for that after. >> >> >> And BTW do not put any comment, it feels much cooler that way. >> >> Stef >> >> >> > |
Re: comments comments comments again
|
|
In reply to this post by Mariano Martinez Peck
So perfect now you have one excuse not to improve.
take two minutes and have a look at the comments of configurationOfSOUP or configurationOfMana. This is the minimum that we can do. You load it and you know to load do X Stef > +99 > Exactly. The few times I needed class comments, the problem was exactly that one: which classes, which where the entry point. > > Now I guess class comments are useful when you're hacking/refactoring. > > > On 27/02/11 19:49, Mariano Martinez Peck wrote: > > > On Sun, Feb 27, 2011 at 4:18 PM, Stéphane Ducasse > <[hidden email] <mailto:[hidden email]>> wrote: > > Hi > > why can we put comment in class comment? > For example I worked on fixing 1.2 and now I want to execute the > configurationOfPharo and no magic invocation.... > JUST A FUCKING EMPTY class comment as usual. > > > So again I feel like an idiot and I'm trying in the dark. You have > NOT A SINGLE REASON NOT TO PUT COMMENT! > One of these days I will hack the system to break the load when a > class as no comment. > > > I was the original author of such class. And yes, I don't write class > comments. I am not use too, usually because: > > - I almost don't find it use > - The browser doesn't help me to see class comments > - Who can guess that under a '?' there are class comments? I discovered > that YEARS after I was already using Smalltalk > - I prefer writting/reading a TestCase for such class (I know it is not > the same as class comments, but it helps). > - I prefer looking at senders/references to such class > - When class comments are long, the browser is FUCKING slow to write > > Even more, in this case, what would you write for example? > ConfigurationOf* are just Metacello Configurations. What would you need > to know about that despite of Metacello? If you know Metacello, you > should know how to query the configuration and get all the things you > want, like groups, packages, etc > > > So now after randomly trying.....I get a nice DNU. > > > But do you think that could be fixed just because of having class comments? > > Apparently my time is not worth anything.... > > So if you want to see if my fix works. Give a try I will not and if > what I did break, may be I will have a look or may be not > because no time for that after. > > > And BTW do not put any comment, it feels much cooler that way. > > Stef > > > > > |
Re: comments comments comments again
|
|
In reply to this post by Stéphane Ducasse
Hi Stéph,
why not implement a tool that takes a package, asks the user to identify the most important class comment, and the creates comments for all uncommented classes in the package that direct the user to the most important class comment? The generated comment could use the squeak code hyperlink support so the user just has to double click in the generated comment to get to the new comment. How about also including in the comment template a suggestion of what kind of entry-point documentation you'd like to see? 2¢ Eliot On Sun, Feb 27, 2011 at 7:18 AM, Stéphane Ducasse <[hidden email]> wrote: > Hi > > why can we put comment in class comment? > For example I worked on fixing 1.2 and now I want to execute the configurationOfPharo and no magic invocation.... > JUST A FUCKING EMPTY class comment as usual. > > > So again I feel like an idiot and I'm trying in the dark. You have NOT A SINGLE REASON NOT TO PUT COMMENT! > One of these days I will hack the system to break the load when a class as no comment. > > So now after randomly trying.....I get a nice DNU. > Apparently my time is not worth anything.... > > So if you want to see if my fix works. Give a try I will not and if what I did break, may be I will have a look or may be not > because no time for that after. > > > And BTW do not put any comment, it feels much cooler that way. > > Stef > > > |
Re: comments comments comments again
|
|
> Hi Stéph,
> > why not implement a tool that takes a package, asks the user to > identify the most important class comment, and the creates comments > for all uncommented classes in the package that direct the user to the > most important class comment? The generated comment could use the > squeak code hyperlink support so the user just has to double click in > the generated comment to get to the new comment. How about also > including in the comment template a suggestion of what kind of > entry-point documentation you'd like to see? I think that it would give the impression that there are comments. Here my point was: "let us think 1 s about what is the key information a user would need? oh of course how to execute the configuration so that it loads" So this line should not be that difficult to add To load me, execute ((Smalltalk globals at: #ConfigurationOfPharo) project version: #stable) load Because I'm lost with latest, default, load .... We should have tool to use class comments to raise their values. I hope soon that we will have a SmalltalkDoc up and running. Stef |
Re: comments comments comments again
|
|
Stef,
Interestingly enough, I have added a class comment to the metacello configuration template class, but when I copy the class, the class comment is not copied ... everything else makes it ... I had hoped that the comment would be present in the copy ... I suppose I could add the comment programmitcally... Dale On Feb 28, 2011, at 12:13 AM, Stéphane Ducasse wrote: >> Hi Stéph, >> >> why not implement a tool that takes a package, asks the user to >> identify the most important class comment, and the creates comments >> for all uncommented classes in the package that direct the user to the >> most important class comment? The generated comment could use the >> squeak code hyperlink support so the user just has to double click in >> the generated comment to get to the new comment. How about also >> including in the comment template a suggestion of what kind of >> entry-point documentation you'd like to see? > > I think that it would give the impression that there are comments. > Here my point was: "let us think 1 s about what is the key information a user would need? > oh of course how to execute the configuration so that it loads" > > So this line should not be that difficult to add > > > To load me, execute > ((Smalltalk globals at: #ConfigurationOfPharo) project version: #stable) load > > Because I'm lost with latest, default, load .... > > We should have tool to use class comments to raise their values. > I hope soon that we will have a SmalltalkDoc up and running. > > > Stef > > |
|
|
In reply to this post by Mariano Martinez Peck
>
> - The browser doesn't help me to see class comments > - Who can guess that under a '?' there are class comments? I discovered that YEARS after I was already using Smalltalk This is a long time I wish we had like in the classic squeak browser a view with class template and class comment... So no need for the ? button... no need to switch (which is disturbing and )... I'd like it but I couldn't get a clue on how to do that in OB 5Ithink this is not trivial) > - I prefer writting/reading a TestCase for such class (I know it is not the same as class comments, but it helps). I also find we could have (automatically) in the class comment pane, all tests that are related to it... with the hyperlink stuff... Last, besides help books which are great, I find sometime a high valuable information is a debug me expression... to understand how something works... That'd be nice to have again, in comments... My 2c Cédrick |
«
Return to Pharo Smalltalk Developers
|
1 view|%1 views
| Free forum by Nabble | Edit this page |