[CLOSED] Implement Node.js for Linux

[core-ai-bot]

Issue by jasonsanjose
Tuesday Aug 20, 2013 at 19:21 GMT
Originally opened as adobe#306

---

Based on an earlier pull request from @timburgess, #278. This also includes commits from #305 that fix the installer build. Finally, I've updated the gyp and grunt scripts to include node.

Original description:

This is some initial work in getting the node process up & running in Linux.

With this code, brackets-shell kicks off a mutexed launch thread that starts the Node executable in a subprocess.
As per the latest linux setup script, Node is launched from brackets-shell/deps/node/bin/node.

A file descriptor in the startup thread is piped to STDIN in the subprocess.
I have experimented with passing input to the node process to eval(). This works but Node appears to buffer it's output if it detects that STDIN is not a terminal. If I run node in interactive mode i.e. 'node -i' it stops buffering but produces a lot more output than I think is wanted.

I'd appreciate getting some feedback on how to run Node as then I can add code to setup a pipe from the subprocess STDOUT to a thread that reads the output.

---

jasonsanjose included the following code: https://github.com/adobe/brackets-shell/pull/306/commits

[core-ai-bot]

Comment by timburgess
Tuesday Aug 20, 2013 at 22:07 GMT

---

+1 to merge. Grunt build produces a binary which runs for me OK and seems to have no issue with node-side extension code.

[core-ai-bot]

Comment by jasonsanjose
Tuesday Aug 20, 2013 at 22:43 GMT

---

Thanks @timburgess. We still need to wait for #305 to land. I'll merge this pull soon after.

[core-ai-bot]

Comment by JeffryBooher
Wednesday Aug 21, 2013 at 18:40 GMT

---

I was able to build Brackets with this branch and the -3 is definitely gone! 👍

[core-ai-bot]

Comment by jasonsanjose
Thursday Aug 22, 2013 at 04:45 GMT

---

Merging. Awesome work @timburgess! Also thanks for reviewing my gyp and grunt changes. Thanks @JeffryBooher for confirming the bug fix.

This. Is. Big. 🙌

[core-ai-bot]

Comment by timburgess
Thursday Aug 22, 2013 at 05:43 GMT

---

No problem. Glad it is now in master and hopefully should kill a lot of the recent Linux functionality questions..

[core-ai-bot]

Comment by timburgess
Friday Sep 06, 2013 at 10:47 GMT

---

@jasonsanjose I'd value being listed in the Sprint 30 Release Notes as a contributor ;-)
https://github.com/adobe/brackets-shell/commits/master/appshell/appshell_node_process_linux.cpp

[core-ai-bot]

Comment by jasonsanjose
Friday Sep 06, 2013 at 16:05 GMT

---

Yikes! I guess I missed this one because I re-submitted the pull request. Sorry about that! I've updated the release notes https://github.com/adobe/brackets/wiki/Release-Notes:-Sprint-30.

[core-ai-bot]

Comment by njx
Friday Sep 06, 2013 at 17:27 GMT

---

FYI, if what happened is that you did a merge/squash and so lost the original commit, in future you can specify --author=<author> to git commit to set the original author. The <author> can be a portion of the author's name or email address.

[core-ai-bot]

Comment by jasonsanjose
Friday Sep 06, 2013 at 17:30 GMT

---

Nope, wasn't a squash. I think we normally just look at the pull request author vs. the commit author.

[core-ai-bot]

Comment by njx
Friday Sep 06, 2013 at 18:01 GMT

---

Ah. Maybe when reopening a pull request, we could put the original author in the title of the new one just so we'll notice it when scanning for the release notes.

[core-ai-bot]

Comment by jasonsanjose
Friday Sep 06, 2013 at 18:15 GMT

---

Good idea!