July 2016 - Pharo-users - lists.pharo.org

System Browser "Implementors of..." drill down?

by Torsten Bergmann

Norbert Fortelny wrote: >Something I have not figured out how to do easily in the Pharo system >browser: I am looking at a method and want to see how a message that is used in this >method is implemented. Have a look at my Message Flow browser: http://forum.world.st/ANN-MessageFlowBrowser-td4849049.html https://www.youtube.com/watch?v=DRd_bzGocQg It should do the job you need. You can load it from Catalog. It mimics a "dive in like" message browser that is found in SmalltalkMT but for Pharo. HernÃ¡n Morales Durand >This was actually implemented in older versions of Pharo (or Squeak), the >browser also included very handy buttons in the middle to save precious >clicks. Yes - this was removed from Pharo but was useful. Stef wrote: >I would also love to scope and manipulate the scope of the list. For >example I would like Some Smalltalks had "Senders"/"Implementors" and "Local Senders"/"Local Implementors" just checking the hierarchy upwards. This was useful and we possibly should include such facilities (scoped to hierarchy, package, ...) in the future again. Thanks Torsten

July 26, 2016

Re: [Pharo-users] System Browser "Implementors of..." drill down?

by Thierry Goubier

I prefer direct manipulation: - shorten the menu in a code pane, add implementors based on the selected / where cursor is in text (aka smart suggestion): below is the menu for the cursor inside #import:provides:. [image: Images intÃ©grÃ©es 1] 2016-07-26 1:33 GMT+02:00 HernÃ¡n Morales Durand <hernan.morales(a)gmail.com>: > See attachment > > HernÃ¡n > > > 2016-07-24 18:19 GMT-03:00 stepharo <stepharo(a)free.fr>: > >> Hi hernan >> >> Could you send a screenshot so that I understand precisely what you mean? >> >> I tend to listen to grumbling people like you and me :) >> >> Stef >> >> Le 22/7/16 Ã 09:24, HernÃ¡n Morales Durand a Ã©crit : >> >> This was actually implemented in older versions of Pharo (or Squeak), the >> browser also included very handy buttons in the middle to save precious >> clicks. >> >> The right thing to do is to replace the "..." non-sense with the selected >> method selector, and to offer another menu item which displays all method >> selectors of the currently selected method, and let the user to decide >> which one to browse. >> >> HernÃ¡n >> >> 2016-07-21 14:54 GMT-03:00 Norbert Fortelny <norbert.fortelny(a)gmail.com>: >> >>> Something I have not figured out how to do easily in the Pharo system >>> browser: >>> I am looking at a method and want to see how a message that is used in >>> this >>> method is implemented. In other Smalltalks (e.g. Cincoms) I use the menu >>> item "Implementors of..." and get a list of all the messages that are >>> used >>> in this method and can choose one to drill down. >>> In Pharo I get an immediate answer but only the implementors of my >>> original >>> method. >>> Why are there those "..." at the end of this menu item indicating some >>> further choice when there is none? >>> How to do this drilling down in Pharo? >>> >>> >>> >>> >>> -- >>> View this message in context: >>> http://forum.world.st/System-Browser-Implementors-of-drill-down-tp4907380.h… >>> Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com. >>> >>> >> >> >

July 26, 2016

Re: [Pharo-users] System Browser "Implementors of..." drill down?

by Thierry Goubier

2016-07-24 23:21 GMT+02:00 stepharo <stepharo(a)free.fr>: > Feel free to improve the tools. > > This is on my list to improve > > - method categorizer > > - finder to use lists instead of trees (because it is way better) > > - message browser (to make sure that from a scope browser, the env is > passed and we do not end up browsing the complete env > > filter based on packages...) > I've already done the last two in my environment, and got rid of the first one (replaced the finder result UI by a scoped browser). I can tell you, this is really fun to use... Except that now, I'm unable to use a classical (Nautilus or OmniBrowser) system browser anymore :( Thierry > But I'm doing too many things. > > > Stef > > Le 22/7/16 Ã 09:47, Markus Stumptner a Ã©crit : > > I agree - the keyboard shortcuts will work for the initiated, but are not > a replacement for the obvious menu entries that even experienced > Smalltalkers from other versions will be missing in Pharo. > > I think it is important to keep the entry barriers low in terms of not > hiding useful functionality that people will be used to from other > environments. > > Markus > > On 22/07/2016 16:54, HernÃ¡n Morales Durand wrote: > > This was actually implemented in older versions of Pharo (or Squeak), the > browser also included very handy buttons in the middle to save precious > clicks. > > The right thing to do is to replace the "..." non-sense with the selected > method selector, and to offer another menu item which displays all method > selectors of the currently selected method, and let the user to decide > which one to browse. > > HernÃ¡n > > 2016-07-21 14:54 GMT-03:00 Norbert Fortelny <norbert.fortelny(a)gmail.com>: > >> Something I have not figured out how to do easily in the Pharo system >> browser: >> I am looking at a method and want to see how a message that is used in >> this >> method is implemented. In other Smalltalks (e.g. Cincoms) I use the menu >> item "Implementors of..." and get a list of all the messages that are used >> in this method and can choose one to drill down. >> In Pharo I get an immediate answer but only the implementors of my >> original >> method. >> Why are there those "..." at the end of this menu item indicating some >> further choice when there is none? >> How to do this drilling down in Pharo? >> >> >> >> >> -- >> View this message in context: >> http://forum.world.st/System-Browser-Implementors-of-drill-down-tp4907380.h… >> Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com. >> >> > > >

July 26, 2016

Re: [Pharo-users] System Browser "Implementors of..." drill down?

by Norbert Fortelny

What about just adding the method itself as the first item to the list. Like it is in VW Smalltalk: attached. Could I try my hand on this? Does anyone know about possible pitfalls to be aware of? Old hands: any thoughts? On 26/07/2016 01:50, hernanmd [via Smalltalk] wrote: > See attachment > > HernÃ¡n > > > 2016-07-24 18:19 GMT-03:00 stepharo <[hidden email] > </user/SendEmail.jtp?type=node&node=4907915&i=0>>: > > Hi hernan > > Could you send a screenshot so that I understand precisely what you > mean? > > I tend to listen to grumbling people like you and me :) > > Stef > > > Le 22/7/16 Ã 09:24, HernÃ¡n Morales Durand a Ã©crit : >> This was actually implemented in older versions of Pharo (or >> Squeak), the browser also included very handy buttons in the >> middle to save precious clicks. >> >> The right thing to do is to replace the "..." non-sense with the >> selected method selector, and to offer another menu item which >> displays all method selectors of the currently selected method, >> and let the user to decide which one to browse. >> >> HernÃ¡n >> >> 2016-07-21 14:54 GMT-03:00 Norbert Fortelny <[hidden email] >> </user/SendEmail.jtp?type=node&node=4907915&i=1>>: >> >> Something I have not figured out how to do easily in the Pharo >> system >> browser: >> I am looking at a method and want to see how a message that is >> used in this >> method is implemented. In other Smalltalks (e.g. Cincoms) I >> use the menu >> item "Implementors of..." and get a list of all the messages >> that are used >> in this method and can choose one to drill down. >> In Pharo I get an immediate answer but only the implementors >> of my original >> method. >> Why are there those "..." at the end of this menu item >> indicating some >> further choice when there is none? >> How to do this drilling down in Pharo? >> >> >> >> >> -- >> View this message in context: >> http://forum.world.st/System-Browser-Implementors-of-drill-down-tp4907380.h… >> Sent from the Pharo Smalltalk Users mailing list archive at >> Nabble.com. >> >> > > > > *CHB1.jpg* (134K) Download Attachment > <http://forum.world.st/attachment/4907915/0/CHB1.jpg> > > > ------------------------------------------------------------------------ > If you reply to this email, your message will be added to the discussion > below: > http://forum.world.st/System-Browser-Implementors-of-drill-down-tp4907380p4… > > To unsubscribe from System Browser "Implementors of..." drill down?, > click here > <http://forum.world.st/template/NamlServlet.jtp?macro=unsubscribe_by_code&no…>. > NAML > <http://forum.world.st/template/NamlServlet.jtp?macro=macro_viewer&id=instan…> > Senders in VW ST.tiff (1M) <http://forum.world.st/attachment/4907955/0/Senders%20in%20VW%20ST.tiff> -- View this message in context: http://forum.world.st/System-Browser-Implementors-of-drill-down-tp4907380p4… Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com.

July 26, 2016

Re: [Pharo-users] Pharo and GIT

by Thierry Goubier

2016-07-26 9:01 GMT+02:00 Serge Stinckwich <serge.stinckwich(a)gmail.com>: > On Tue, Jul 26, 2016 at 12:33 AM, Sean Glazier <sglazier456(a)gmail.com> > wrote: > > Thank you, Peter for documenting that :-). While I think git tools are > OK, > > my opinion though is that Smalltalk hub should be moved forward as well. > At > > cincom the Store experience started out painful and it has its quirks, > but > > one of the advantages is that I could write tools to do things like > browse > > senders and implementors in the repository. I git you are working with > files > > and text and it does not have the notion of classes and methods and the > > value of being able to see the history of the class / methods. > > This is possible to see all the history of a given method on github: > > > https://github.com/pharo-project/pharo-core/commits/6.0/CodeImportCommandLi… And it can be extended to see the history of a class (track the relevant directory) or project (same). FileTree obviously remaps Smalltalk code concepts (which are already quite simple: class, method) into a directory/files structure, making the conversion easy. Only people who insist on storing a full class or package in a single file are purposely destroying that mapping. Note, for those who miss VW Store, that Squeak has a Magma mapping for code. Note that Pharo had that ability to query the git history for a method since I implemented it a few years ago... Thierry > > > -- > Serge Stinckwich > UCBN & UMI UMMISCO 209 (IRD/UPMC) > Every DSL ends up being Smalltalk > http://www.doesnotunderstand.org/ > >

July 26, 2016

Re: [Pharo-users] Pharo and GIT

by Serge Stinckwich

On Tue, Jul 26, 2016 at 12:33 AM, Sean Glazier <sglazier456(a)gmail.com> wrote: > Thank you, Peter for documenting that :-). While I think git tools are OK, > my opinion though is that Smalltalk hub should be moved forward as well. At > cincom the Store experience started out painful and it has its quirks, but > one of the advantages is that I could write tools to do things like browse > senders and implementors in the repository. I git you are working with files > and text and it does not have the notion of classes and methods and the > value of being able to see the history of the class / methods. This is possible to see all the history of a given method on github: https://github.com/pharo-project/pharo-core/commits/6.0/CodeImportCommandLi… -- Serge Stinckwich UCBN & UMI UMMISCO 209 (IRD/UPMC) Every DSL ends up being Smalltalk http://www.doesnotunderstand.org/

July 26, 2016

Re: [Pharo-users] Pharo and GIT

by Peter Uhnak

On Mon, Jul 25, 2016 at 08:08:49PM -0300, HernÃ¡n Morales Durand wrote: > 2016-07-25 19:33 GMT-03:00 Sean Glazier <sglazier456(a)gmail.com>: > > > Thank you, Peter for documenting that :-). While I think git tools are OK, > > my opinion though is that Smalltalk hub should be moved forward as well. At > > cincom the Store experience started out painful and it has its quirks, but > > one of the advantages is that I could write tools to do things like browse > > senders and implementors in the repository. I git you are working with > > files and text and it does not have the notion of classes and methods and > > the value of being able to see the history of the class / methods. It is > > valuable when needing to understand not just the current implementation but > > where it came from. In Store, I could search and see when a method was > > present of a class and in what version lets say it got dropped > > unintentionally for instance. Even more important I think is to also note > > thing like renames and in cincom namespace renames and moves. > > I was working a while back on a way to have a persistent diary for a > > class that could note these histories and maintained and more importantly > > searchable from the image. > > I wrote tools too, to attach to a number of databases (repositories) and > > search for classes and the comments. Helpful when you are searching to see > > if someone else has solved a problem before. Git tools can tell you a lot > > about the repository you are publishing to and comparing code etc. But it > > does not help you to search across repositories and the data. > > > > I know this idea is a tall order. If we improved our tools to beable do > > this, no matter what the repository is behind it, that would be very > > helpful and powerful. I think it is a challenge in git because it is > > dealing with text and does not have a notion of what a class is etc. If we > > continue and put in more powerful search abilities, it will require using > > different paradigm in representing code so we can do more powerful things > > with the repository. We can browse the class without loading it but that > > gets us only so far. > > > > > That is exactly the core problem with Git. It was conceived by people very > used to think in terms of files and directories. Because that is how the system (linux) works, so it makes sense that it works that way. > And it was promoted as a revolutionary tool. It is. > If the model of Git were well-done would be adaptable to handle references > of classes, and methods, of any typical programming language, regardless of > paradigms. Git is used also for other things, not just programming languages. Peter

July 26, 2016

Re: [Pharo-users] Pharo and GIT

by Peter Uhnak

On Mon, Jul 25, 2016 at 06:33:48PM -0400, Sean Glazier wrote: > Thank you, Peter for documenting that :-). While I think git tools are OK, > my opinion though is that Smalltalk hub should be moved forward as well. While SmalltalkHub works for code storage, it is also just that, a storage. Trying to run a service competing with the likes of GitHub, Bitbucket, or GitLab is frankly waste of resources that could be better invested into things that are relevant for the mission of Pharo. > I git you are working with files and text and it does not have the notion of classes and methods and > the value of being able to see the history of the class / methods. I've seen this argument many times and yet it is completely false. You can actually retrieve the history of a single method even for languages that truly do use just files (C, ruby, â¦), although a fair bit of scripting would be required, but the history is there. The used FileTree format that writes each method to a file makes this much more simpler. > thing like renames and in cincom namespace renames and moves. Funny that you mention renames, considering that SmalltalkHub cannot handle renames or track changes/moves across packages. > I know this idea is a tall order. If we improved our tools to beable do > this, no matter what the repository is behind it, that would be very > helpful and powerful. I think it is a challenge in git because it is > dealing with text and does not have a notion of what a class is etc. class -> directory, method -> file makes it much more easier. But even if it was all a dump into a single file, the bottom line is, it is always only a question of tooling. After all, don't forget that whatever format you use, if you are storing it on hard drive it's in a file. You just have tools on top. Peter

July 26, 2016

Re: [Pharo-users] Pharo and GIT

by Dale Henrichs

On 07/25/2016 03:33 PM, Sean Glazier wrote: > I git you are working with files and text and it does not have the > notion of classes and methods and the value of being able to see the > history of the class / methods. FileTree is the layer that lives between the directory structure on disk and the image ... FileTree interprets the directories and files on disk and creates a snapshot composed of definitions for classes, methods, etc. Saying that git doesn't have a concept of classes and methods is equivalent to saying that postrgres (the underlying database engine for Store i believe) doesn't have the concept of classes and methods :) It is the tools that matter, not how the bits are written to disk... Since FileTree stores each method as a file, each method does have history, as does each of the parent directories (class, package, project) ... What is true is that as the tools stand today it isn't possible to access "method history from git", but it is a tools issue... > It is valuable when needing to understand not just the current > implementation but where it came from. In Store, I could search and > see when a method was present of a class and in what version lets say > it got dropped unintentionally for instance. Even more important I > think is to also note thing like renames and in cincom namespace > renames and moves. All of this is available from git given the proper tooling ... Dale

July 25, 2016

Re: [Pharo-users] Pharo and GIT

by Nicolas Passerini

> > > Well I wasn't sure if I should be reporting things now (because obviously > it's incomplete), and I saw 58 issues on github, so didn't want to add > more. > But if you want some feedback :) > Yes, feedback is necessary... even knowing that you are using it is good news for me. > * (confusing) The colors of the working copy / local repo should > be switched, because now it's showing me new code in red, and removed code > in green, which is confusing > Yes, I know. I was stubborn in putting my version in the left, but I didn't find the way to customize the diff tool to show colors properly. Also the diff does not highlight correctly, at some point I will have to get my hands on it, but I didn't have the time yet. > * (really minor) Opening the synchronizing window by double > clicking on the repository (instead of clicking on the button) will open it > four times > Again, yes... I do not know if this is a bug of Glamour, I've posted a question but didn't get answers... I have to insist :) > * (docs) from readme file it's not clear whether I have to run the > update command; running it will brick the installation > Update is not mandatory, but should be a useful way to get the latest version. I'm not sure of having understood you, is that you tried it and it failed? > * (yay!) I am really happy I can use it side-by-side with > gitfiletree so I always have a fallback option > :+1: > * (unsure) I wasn't sure from the docs / experiments whether I can > point it to local clone, or always have to use github > Yes you can. When creating a repository, just change the default location. (Yet it is a bit confusing because it still asks you for a remote url instead of using that of the clone you selected... I have to improve it.) > I am really looking forward to this project, and I am slowly starting to > use it for some smaller projects, so more feedback will certainly come. :) > > Thanks! > > Peter > > Thank you!

July 25, 2016