Hello, I was looking around for a project in Github and find the project "Implement scripts to generate source tarball of Atom and its sibling projects" very interesting. Also most of the Skills require matches to my skill sets. I've also participate in last year's GSoC.
I just want to clarify one thing. By now what we have a .deb for Atom which is compiled binary source. Now what we require is the source tarball for Atom so that ultimately will help us creating our own PPA for Atom.
I've been researching around for the way to proceed this problem, but it will be good if I can get some starters with the problem.

One more idea which I was thinking can be to implemented Atom is to introduce Chef based Cookbook.
It will be an enhancement for developers who are trying to make contribution by introducing vagrant environment for development process.
The benefit of this would be that developers can create their development environments from the same configuration, so whether working on Linux, Mac OS X, or Windows, all of them can run code in the same environment, against the same dependencies. Vagrant provides a disposable environment and consistent workflow for developing and testing. One can quickly test things like shell scripts. Vagrant also automatically set everything up that is required for that app in order to focus on doing what you do.
The other thing that needs to handle in parallel to vagrant setup is provisioning. Provisioning in Vagrant allows to automatically install software, alter configuration and manage things. The provisioning method which can be implemented is Chef. Chef operates using cookbooks, i.e. self-contained scenarios such as “installing and configuring required dependencies”. Environments are assigned cookbooks and Chef makes sure that everything is available and up to date, thus providing us the automation and re-usability part of the recipe.
This is the part where I'm experienced working last summer at GSoC.

Please let me know is this implementation worth it or not.
Thank you.

Based on the discussion in #15 on command line based tools for atom-shell, I would like to propose a project idea to build a workshopper for getting started with atom-shell development.

A workshopper is an interactive tutorial inside the command line based on a specific topic, that contains a tutorial followed by a problem on the topic that the player tries to solve based on what he/she learnt in that tutorial.

Workshoppers were started by the nodeschool community, and today we have workshoppers on a wide range of topics, from the basics of JavaScript and node, individual topics like streams and promises, and for learning software packages like express, browserify and leveldb. GitHub's own @jlord maintains the
official workshopper on git.

During the summer I would like to create a workshopper for atom shell. A rough plan is to start with
installation, and finish with packaging and distribution strategies, while in the middle lessons we can focus on topics like App Menus, Task Tray support, Events(Online/Offline), kiosk mode, protocol handling etc. We can include lessons on the basics of browser and client scripts and the separation between them as well as a portion on selenium and webdriver and how developers can run automated tests for their atom-shell apps. This is obviously a rough idea for the project, just off the top of my head.

A rough schedule for the project would involve designing 1-2 lessons a week, with a tutorial and appropriate documentation, and I would also like to do a preliminary release at the time of the mid-term evaluation for the community to try it out and provide feedback.

Oh and my name is Suyash and I am an undergrad at IIT Roorkee

Cheers!

Hi. I'd like to participate in improving Atom Shell. This idea is very interesting for me. What can I do to be selected on this task? Thanks.

Hi guys,

I find this issue interesting for me cause I'm not in latin group too and it skills required are quite close to me.

Cheers,
Pavel

Getting rid of the XCode Command Line tools dependency would be really useful for the Boxen Project.

Are there any other resources for learning about the command line tools other than this one.

/cc @MikeMcQuaid

Hi. I am Huang Yue from the School of Computing, National University of Singapore. I am now doing my bachelor degree in Computer Science and I am 'somewhere between second and third year' since I am trying to graduate in 3 and a half year.

I am particularly interested in the proposal of about prioritization of taps.

I have used Homebrew for a while and I have read a portion of its code. As far as I am concerned, implementing this will actually make a great change in Homebrew's structure.

So far I realized that currently we do not actually have 'meta information' of taps and thus they cannot be managed efficiently. I suppose that the expected outcome of this proposal is actually to implement a better structure for taps so they work more similar to 'sources' in other package managers like apt-get or yum. (And after that, allowing prioritization is easy, much like how yum-plugin-priorities works.)

@MikeMcQuaid Am I thinking the problem correctly? If so I am probably going to design and propose an architecture of a new version of taps, as well as analyzing how much changes I need to make. Thank you very much for your help!

Hi,

Skimming through the tokenizer source and the summary, I gather you guys would like to introduce a sense of grammar in the lexical phase to create some sense of what the language the input file is in before you pass it's tokens through the bayesian classifier? Or additionally also determine whether a particular token would be of interest in further stages of the classification, i.e. pass only particular "tokens of interest" to the next phase, by also taking into account the syntax of the language?

I am not able to fully understand what the goal would be here and would really appreciate it if you could clarify whether I'm on the right track.

Also, any other specifics you guys had in mind regarding the way you want this to be implemented is appreciated as well, @arfon @bkeepers

I've undertaken courses in language processing at uni and would really like to put them to use here (I'm interested).

P.S. : If this is indeed what we have set out to achieve, I have some other ideas I'd like to discuss, perhaps via e-mail.

@kevinsawicki since this seems to be your area
Do you have any recommendations for debuggers to start stepping through the code to figure out where the issues might be?

Hello. I am Hasit Mistry and I am doing my MS in Computer Science and Software Engineering from University of Washington Bothell.

I have been using fish for over a year now and I was a zsh user before that. I have been tinkering with the autocompletion of hub for some days and I have noticed the following:

1. fish autocomplete script file is missing. Is that something that I can start working on? A little help would be great.
2. I recently learnt that fish can generate completion scripts using man pages. I posted a comment on hub repo issue #656 asking for help and providing a suggestion.

I am interested in implementing a way to define these shell completions for bash, zsh and fish. While fish does this automatically from man pages, hub expects an alias to git and due to this reason, there are no options in the hub man pages for fish to generate an autocompletion script from. I confirmed that by looking into the ~/.config/fish/generated_completions/ folder.

A possible way of doing this:
Define all the completion options in a text format (JSON/XML maybe?) and make a Go program to read this file and generate autocomplete scripts for each shell. A Go script along the lines of fish_update_completions script.

I see @mislav is in charge of mentoring Hub as suggested by @johndbritton. I am very excited about taking up this task and any help or pointers in the right direction would be of great help.
Thank you.

Hi guys !
I was wondering if you had any projects including iOS development on your mind ?

Hello. I'm a college student from China and very interested in contributing for homebrew in GSoC.

However, I haven't got the meaning why we need to package all the dependencies in one directory? An important reason why we use a package manager rather than manually compile software is that it can deal with dependency automatically. It won't trouble users if mysql and its dependency openssl are all bottled.

(other than #github, of course)

I'm quite interested, but I'd love to speak to some of the mentors in more detail about multiple ideas.

Hello everyone
I am reading Linguist code and I am getting little familiar with it. Can anyone please tell me now what should I do if I want to fix a misclassified Language. It would be great if you could please mention which are files should I read for that.
Thank You
Pawandeep Singh

Hi @kevinsawicki ,

I am Vishnu Teja doing my B.Tech from a deemed University National Institute of Technology, Surat, India. I am would like to do a project in Github through GSoC 2015. My skills are JS, HTML, CSS, Unit testing and Integration tests. I have been contributing Mozilla's product Firefox OS {JS, HTML,CSS} for a long time. I would like to do projects based on JS which are mentioned in idea's list which are based on Atom. Please help me with some pointers from where I can start fixing some bugs in code base to get myself fluent to codebase. I tried pinging you on IRC at #github And also please help me where I can ping you on IRC.

Thank you :)

Hello, I am still in high school and turning 18 only on August this year. I would love to take part in Google Summer of Code this year, isn't that possible? Thanks in advance.

Hi,

I'm interested in the GSoC projects with Atom, especially the one relating to improving the performance of the parser. I was wondering if @kevinsawicki had any advice about where to start. I'm experienced with C++ and I've worked with javascript in a couple of projects, but I haven't done any work with the atom source. I'm getting the build environment set up, and I've been looking at atom/atom#979 which looks like it would be a good starting point. I'm also hoping it will be a bit easier.

I appreciate any help you can offer. Thanks.

Hi, I have roughly viewed the code base of linguist project. Before starting to play with the code, I would like to ask several questions.

1. How is the performance of Linguist in production? If someone starts work on this project, will he/she has access to the records of GitHub users reporting misdetection on their source files?
2. How are the sample data selected? This is quite critical to the training models of bayesian learning.
3. One of the project ideas for Linguist project is to aggregate the predictions from multiple predictors. However, I didn't really see how you current aggregation strategy is from your code base, i.e. how do you detect the programming language using the multiple classifier at this moment? Sorry if I missed something.
4. Have you ever thought about using classification algorithms other than bayesian classifier? According to the paper in reference, bayesian classifier looks like a very good fit for this problem. However, I still wonder how other classification algorithms works on programming language classification problem.

One intuitive attempt to the second project idea is to apply several different classifiers, then we can maybe use some boosting techniques to aggregate them. However, the deterministic detection strategies (by looking at file name extension, shebang, etc) should be very powerful (predict correctly almost every time), and will take much larger weights comparing to the probabilistic classifiers. I am very curious how it will actually work on real dataset.

Thanks!

I've been going over the discussion in atom/atom#979 and looking at what's causing the performance problems when it comes to long lines. On the surface the problem is that Atom treats lines a single strings, which means that a long line can be subjected to multiple O(n^2) regex searches in the process of parsing it, which takes huge amounts of time and results in hangs. @maxbrunsfeld has suggested replacing the regex searches with a single linear time scanner, and @prasoon2211 has suggested replacing the existing regex library with one that performs in linear time. While these would alleviate the issues with parsing times, I think there are more core issues with the way that Atom represents documents that can be fixed to yield more overall improvements.

My proposal is to further break up individual lines into smaller segments that allow editing and parsing to take place at a more manageable scale. This would allow an effective cap on how much text would need to be parsed in a single call, which allows for better scaling even with exponential regex search times. I think this solution is a good start towards a more efficient internal representation of documents for Atom. It should be fairly easy to implement in CoffeeScript without having to drop to native code, and even if it fails to yield a significant performance increase it should be fairly illuminating as to where other improvements can be made.

As for GSoC, I'm guessing this alone will not be sufficient to fill the work period, so I think the next step would be to re-implement the most performance-heavy parts of the internal text buffer in C++. This would allow directly interfacing with the regex parser without the need to handle strings in javascript, and would reduce the burden on the garbage collector since string processing tends to be pretty rough on the GC (though perhaps this is not as much of an issue with Atom).

I'm open to all thoughts and feedback. Thanks!

/cc @kevinsawicki @bkeepers

Most gsoc orgs allow students to come up with their own project ideas, that should be innovative but should also be restricted to specific domains. Who should prospective applicants get in touch with to discuss their ideas, and exactly which GitHub open source projects are within scope for such ideas ?

I've been using teachers_pet this semester for grading, and have had my co graders using it as well.

There are some features I would like to add, but I've gotten feedback that there are people would like the option of using a GUI or the CLI.

I'd like to get some thoughts on this.

/cc @johndbritton

@arfon @bkeepers I wanted to research and work on this projects, could you guys give me some pointers or guide me as in where to get started

Hey ,
I just had an idea for a gsoc project and I'd like to share. The command line interface of git may seem daunting to new contributors which results in not contributing to the code world at all. What I'd like to suggest is a project (a simple one) in which we just have to create a gui for git.

Thank you,

p.s: I'd like to implement this using Tkinter python :D Or swings java

Hi dear mentors,

I'd like to know is there any special requirements for the proposal document submitted.
For example:

1. The format of the document (plain text, docx or pdf)?
2. What information should be included about the candidates?
3. Instructions for naming the document that will be submitted?

I suggest to add in atom autocompletion for JS based on static analysis. Now atom-completion collect parsed identifiers:

But there are already exists tools (like babel) with great api for retrieving js ast.
So we just need to write tool that gives more brains to completion mechanism.

Maybe, if it's possible do during summer - parse dependencies and add correct completion for them too.

Did you thought about Atom as an IDE? I mean Atom supports basic completion based on tokens inside file, but it's not a big deal to parse required files (in js).
Also atom natively supports js (cause it's written using it), so we could develop something like spy-js in webstorm. We could collect runtime information for completion improvement.

I have approximately 7 years of experience in using C++ language and many of the libraries made for it. I have also developed softwares & games using Win32 API, Qt 4, SDL, OpenGL. But, I have NOT done any professional course in software engineering. Currently, i am persuing B.Tech in Electrical Engineering.
For GSoC, I am interested in the project -
Improve printing support of Atom Shell.
.
I wanted to know if my degree will be any hindrance in my selections for the Project.
Also i have NOT contributed to any open source projects till now (Although i have started trying that in past 1 month).
Any suggestions in this regard would be helpful!
.
Regards,
Abhinav

The project idea sounds interesting. I looked at the relevant issues which were updated today in the README file but can I get some more additional details about it? The "summary" seems to be too less about the nature of the project.

Since the official application period opens on 16th, will GitHub open an official medium for students to get their proposal reviewed for completion before students submit their proposal to Google?

This is important as GitHub hasn't provided an official template for proposals, so ideally it would help if students had a medium to make sure that their proposal was complete, in the sense that all information that GitHub expects in a good proposal is there, as well as potential mentors can provide some valuable feedback on any additional information that students can add to their proposal, and increase their chances of getting selected for GitHub.

If not, is it advised for students to email potential mentors their proposals, and get valid criticism and feedback, @MikeMcQuaid agreed to this in #46, which is good news... for Homebrew, maybe someone from atom, linguist and hub teams can also agree to review proposals.

Cheers!

/cc @zcbenz @kevinsawicki @jlord @bkeepers @mislav

After skimmed over the source code of go-octokit and some related material, I found the key problem makes the client not eligible may stands in Golang's characteristic of static typing.

Lack of meta programing seems not only lead to code redundancy, but also prevents the implementation of a general library that can be used for other APIs.

But as soon as I see the word hypermedia, I thought of a very similar thing: hyper json schema (a good article about it) . Tool chains around hyper json schema are rich and it also has some abilities to define nested restful resource.

In fact, there already exists a Go library that translates hyper json schema into API client in Golang: Schematic. It has been used by the heroku's fast CLI client hk as the generator of API agent.

I guess Schematic may not be a good solution for Hub because of the following problems:

1. Hyperlinks is predefined in hyper json schema other than response to what the API gives, so it is not as flexible
2. The protocol generated is not compatible with go-octokit and migration means a lot of modification of Hub
3. Lack of API discoverability

But I still regard hyper json schema a good choice and maybe generated go code (with some tweak) is better than current hand writing redundantly code.

So, is it acceptable to transform go-octokit from human writing to generated? Maybe at first a generated client with exactly same interface and generally changed to one with more elegant protocol along with Hub?

BTW: I still have some other elementary ideas but I'm not sure is it suitable to discuss them in a public place like issues. Maybe email in private is better?

Hi @kevinsawicki
I want to work on this project under github umbrella. I am already done with building environment and currently i am contributing to atom packages and have already submitted patches .Also i got aware of key bindings and other important things require for working on packages and themes.
I need your guidance to work on this project, so that i can prepare a feasible and good proposal for it. I successfully completed GSOC -14 under mozilla organisation last year. My work is mostly on HTML, css, javascript, node, unit testing (spec) using jasmine, mocha frameworks and coffeescript.
Thanks

I like this feature which is offered by JetBrains PHP IDE. It helps to easily deploy your website and also keep your code base in sync - but so does git along with a version control system but this feature can be very useful, your views/suggestions?

Also there's a large scope for command line integration for directly compiling and running the code. In the summer I'd like to work on these ideas.

I think it would be a great feature if homebrew supports brew install ruby --version=2.2.0.

I find that homebrew once supported brew versions to list all versions of a formula so that you can manually checkout and install a special version. And I'm wondering why it's removed now.

@MikeMcQuaid I am very interested in the Homebrew's tasks, but I have a question about "Handle formula renames and deletions": is it necessary to use Git to handle the renaming or is feasible to use some kind of migration file that keeps track of formulae renaming?

As I was reading the idea list of Linguist I was not sure whether I am interpreting the exact meaning of both the ideas. This is what I interpreted of both the ideas.
--- Use grammar bundles to tokenize languages ---

In this we want to first detect the language and then highlight the syntax of the corresponding code .

--- Weighted scoring for detection strategies ----

In this we want to improve the existing code of language detection.

It would we very thankful of you if you could please correct me if I am wrong in understanding these ideas.

Thank You

I have contacted @zcbenz privately and he suggested me creating an issue here for future questions about this project. I'll ask my questions as comments below. For questions that aren't very related to GSoC, I'll create issues in relevant repos (or topics in discuss.atom.io).

FYI - here's the full quote of his mail:

Subject: Re: About GSoC 2015
From: Cheng Zhao
To: Xue Fuqiao

Hi Fuqiao,

Your contributions to open source are really impressive, and I'm glad to
know you want to write an Atom package!

The simplest idea of IDE/editor plugin to create an Atom Shell app and run
it within a few clicks. And we can go further by adding ability to create
distributions or installers, or providing multiple templates of apps that
users can choose from.

Many users request for a command line tool
<https://github.com/atom/atom-shell/issues/804> to do that, so you can also
write one and then wrap it with editor plugin.

I don't have much requirements on what the editor plugin would be like, or
which editor/IDE you would write plugin for, you have the freedom on most
things.

For how the template app would look like, you can have a try of
atom-shell-starter <https://github.com/atom/atom-shell-starter>, there is
also some interesting discussions there.

For future questions you can create an issue at
https://github.com/github/gsoc, that repo is for GSoC related things, and
other people can also help you there.

Cheng

On Tue, Mar 3, 2015 at 4:52 AM, Xue Fuqiao wrote:

> Hi Cheng,
>
> (If you are more comfortable with Chinese, I can also write in
> simplified Chinese in future mails.)
>
> I am a Chinese student and an Atom user. I'd like to work on Atom (or
> related projects like Atom Shell and apm) under the GitHub umbrella
> for GSoC 2015, especially this project:
>
>     * IDE and editor integrations to develop Atom Shell apps
>     Summary: We would like to make developing Atom Shell apps easier
> by providing integrations with popular IDEs and editors, it should
> able users to create and debug Atom Shell apps without leaving the
> developing environment. Examples are like Atom package, Visual Studio
> extension, and WebStorm plugins.
>     Expected Outcome: A working Atom package or IDE plugin for
> developing Atom Shell apps.
>
> I'm really interested in text editors. I was an Emacs committer[1][2]
> and pretester. I'm also the administrator of EmacsWiki.[3] In
> addition, I contributed to some Emacs packages (such as
> ergoemacs-mode[4]). Besides Emacs and Atom, I have also used editors
> including (but not limited to) vi/Vim, Brackets[5], Sublime Text,
> Zed[6], Bluefish[7], gedit[8], jEdit[9], and KWrite[10] for some time.
>
> I'm familiar with JavaScript and I've been learning/coding JavaScript
> since 2006. Actually, ActionScript (which is a dialect of ECMAScript,
> IIRC) was the first programming language I learned. Moreover, I've
> contributed to some free and open-source JS projects (such as Firefox
> OS), and edited many entries in MDN[11] in recent years. You can also
> see me in the `about:credits` page of Firefox desktop/mobile/OS.
>
> I had previous experience of GSoC, too: I successfully participated in
> GSoC 2013[12] and contributed many patches to Emacs before, during,
> and after the GSoC coding period.
>
> You may also see some of my contributions to FOSS here:
> https://www.openhub.net/accounts/xfq
>
> For this year's GSoC project, I plan to write a working Atom package
> for developing Atom Shell apps.
>
> I like Atom very much and it's my main editor now. I've reported some
> issues[13][14] and contributed some minor patches[15][16] to Atom and
> its related projects. Currently, I'm learning Atom Shell and trying
> building some simple packages for Atom. I'll do some research and
> write a proposal for the editor/IDE integration project in the next
> few days. In the meantime, I'll also keep reading discuss.atom.io and
> join the #atom-shell channel on Freenode.
>
> BTW, I have a question: which kind of editor/IDE integration is needed
> for Atom Shell? Pull `atom-shell` from npm and generate a directory
> like https://github.com/dougnukem/hello-atom? Run Atom Shell apps with
> a single command/click? Or something else? Would you please share your
> conceptions so that we can discuss and research this idea in as much
> detail as possible (such as the scope of the integration, including
> which parts are critical versus optional for the summer timeline).
>
> Can you point me in the right direction? Thank you very much!
>
> [1] https://savannah.gnu.org/users/xfq
> [2]
> http://git.savannah.gnu.org/cgit/emacs.git/tree/etc/AUTHORS?id=emacs-24.4#n4432
> [3] http://www.emacswiki.org/emacs/EmacsWikiAdministrators
> [4] https://github.com/ergoemacs/ergoemacs-mode
> [5] http://brackets.io/
> [6] http://zedapp.org/
> [7] http://bluefish.openoffice.nl/index.html
> [8] https://wiki.gnome.org/Apps/Gedit
> [9] http://www.jedit.org/
> [10] https://www.kde.org/applications/utilities/kwrite/
> [11] https://developer.mozilla.org/en-US/dashboards/revisions?user=xfq
> [12]
> https://www.google-melange.com/gsoc/project/details/google/gsoc2013/xfq/5885170347409408
> [13] https://github.com/atom/atom/issues?q=is%3Aissue+author%3Axfq+
> [14] https://github.com/atom/tree-view/issues?q=is%3Aissue+author%3Axfq+
> [15] https://github.com/atom/atom/pulls?q=is%3Apr+author%3Axfq+
> [16]
> https://github.com/atom/atom-shell/commit/5083a7505dd8d7b1f90aad0a182b80f85e70885d
>

@kevinsawicki

Hi. I went through the issue 979 and left a comment. Please read it (and the various links mentioned within) - there, I've made a case for changing the regex engine. Let me what you think of the idea.

Thanks!

Can I develop it as an eclipse plugin?

I have approximately 7 years of experience in using C++ language and many of the libraries made for it. I have also developed softwares & games using Win32 API, Qt 4, SDL, OpenGL all for my own learning purpose. Currently, i am a 2nd year B.Tech student at IIT Indore,India.
For GSoC, I am interested in the project -
Improve printing support of Atom Shell.
.
Any suggestions in this regard would be helpful!
.
Regards,
Abhinav

Hi, I'm Vlad Shablinsky. I'm studying computer science in Belarusian State University. I really like Homebrew, it’s the project that I truly want to work on.

I have some skills in python, git and unix, but I’m not very experienced in ruby. I know it a bit, but I can’t say I know it very well.

I submitted a pull request with some fixings that brew audit shows and now I want to go further. I'm interested in "Handle formula renames and deletions" project. As far as I understood the project is to provide some kind of brew rename A B that handles all the renamings of binaries, directories and all of this stuff is done locally. So after the renaming I have B and I don't have A and I still have to be able to brew upgrade B like if it was A. Do I understand the task correctly?

Hi @kevinsawicki , I would like to show my interest in participating in GSOC-15 as a student. I am particularly interested in the suggested idea of Implementing support for non-US keyboard layouts in Atom. I have set up a local build of Atom and I am preparing my proposal for it. Could you guide me on what my next step could be?
#35 #37

first I'm not one of GSOC member, but I'm interesting about this.

so how could I start this? or couldn't
what have you done?
what need be done?

@zcbenz thanks...

Hi, I'm interested in participating with Atom for GSoC 2015. I'm an active open source contributor and quite well versed in CoffeeScript, JavaScript, Python, C, C++. I'm interested in the following ideas for Atom:

1. Improve Atom's parser performance
2. Implement support for non-US keyboard layouts in Atom
3. Improve text rendering in Atom

I've already set up my dev environment. @kevinsawicki could you please point me in the right direction as how to proceed and which project would be the most beneficial to the community so that I could start strengthening my proposal?

I would like to discuss a proposal for Homebrew: a mechanism to check whether a formula is outdated with respect to the upstream version.
I have already started a conversation at Homebrew/legacy-homebrew#33433, and I think the GSoC could be an occasion to re-propose this feature. I have already created an external command (homebrew-livecheck) that tries to achieve this, but it is still an alpha version and it is very fragile.

I would like to know if this could be a subject of interest for the Homebrew project.

Hi! I just saw that github was accepted into the GSoC and got excited about working on features I would really love to see on github. These are features I'd like to see on github that I feel I could work on:

• PDF preview mode (with pdf.js)
• LaTeX preview mode
• LaTeX editing mode

I know that github itself is not an open source project, but I guess it could be possible to develop an open source solution to make these features easy to integrate into github.

What do you think about these?

Hello All,

I'm really interested working on the tokenization of grammar bundles for linguist.

Does anyone have any good resources for learning more about language learning and tokenizing strategies? I would like to be able to try and make a more concrete plan in my proposal.

Thanks!

Problem:
A big pain point of developing with a dynamic language is that you need to look into libraries' document to figure out their interfaces and how these interfaces should be used. Even you are familiar with the semantic of those interfaces, you may still need to look into the document to make sure you put arguments in the right order. This is really frustrating. Static analyzing does not always get its job done due to the dynamic nature of the language.

And this problem is more painful in the javascript would because JS supports only minimum language feature and lots of things are handled in runtime by manipulating arguments.

There is already some type inference engine designed to static analyzing JS code and provide completion information like https://github.com/marijnh/tern. But they could be fail to analyze complex library like https://github.com/request/request, due to the super dynamic nature of the language.

Solution:
My solution tries to address down the this problem, by introducing the type definition files of TypeScript as an external, machine understandable source of library document.

Even though you can make extremely irregular interface in JS, programmers are all most always make their interfaces fit into some rules. That's why we could use type system to describe the interface, like what TypeScript trying to do. And the good part of type system is that it can be understood by algorithm. So we could take them as some kind of machine-readable document of the library.

And lots of pure javascript libraries have well-written type definitions in TypeScript, like this repo https://github.com/borisyankov/DefinitelyTyped/ They are created for the TypeScript guys to bring the pure javascript libraries into their type checking world. But we can also use them as documents for javascript developers. By parsing them, we can implement an awesome autocompletion system and library exploration in Atom for JS development.

Expect Output:
A Atom extension that provide a way to give autocompletion and typing information based on TypeScript definitions for javascript development, and also provides auto-completion functionality.

If you have any idea of my proposal, please let me know. I am also open to the listed ideas, like
the parse performance problem,

1. Formulae should be able to interpose the link period, i.e., decide which directory to link or not.
2. Besides uninstall, homebrew should introduce a purge method that removes all files, including some generated files under etc / var and downloaded files under lib.

Hello, I have an good project idea, which may help many developers as well as other organizations, so how do I add that as a project proposal under github ?

Thanks

Hi,
I am quite interested in the project "Create support for defining extensible shell completions" .And i want to do the project under Gsoc'15.

I am quite familiar with bash,and have used fish and zsh for more than 2-3 months, i had previously done some research about bash completion when i was trying to think of implementing fish like autosuggestion in bash.
please give me some direction to proceed for this project,and some reference material to get me started