We used this for a project before, just be aware that the DOM changes regularly for gmail, and if an API you rely on breaks you will be SOL until someone fixes it.
Thats exactly what I was wondering about; you can see in the latest merge 11 days ago that a selector had to be updated with random junk to match a recent change (from '.ii.gt' to '.ii.gt div.a3s.aXjCH')
Agree.
I did a gmail plugin before gmail.js came out.
You kind have the same problem if you are the only one using your custom lib and you dont have a huge army of others who could fix it. And it's… so… much… more exhausting
It seems like you might be able to come up with a way to auto-generate these selectors based on known content on the page. ie, find the thing that has this text in it and use that as the selector from now on.
back then we even ended up using surprisingly complex structures that actually are close to what you describe but clientside. as in: we didnt rely on classnames but often on positions next to some text.
The issue is that gmail will update at anytime, without warning, so there will always be a period when your service will break. And you need to have unit tests running often to figure out when it breaks, and have a dev on call 24/7 to ensure a speedy fix. This is a universal problem with extensions targeting website DOM's, but worse if you have an SaaS product that has an SLA.
Founder of Streak here (we make the InboxSDK) - this is one area where we take a slightly different approach than Gmail.js.
We host our SDK and you remotely load it at runtime. The benefit is that we detect breakages due to gmail and automatically update the SDK. Your users just need to refresh gmail to load your extension and the latest SDK which is compatible with any changes Gmail makes.
InboxSDK looks awesome. I took a look at it several months ago, and we want to build something on GMail. We're a little concerned about the licensing and availability of the library, though. Is it possible that a month from now you decide that we're competing with something you want to do, and stop serving the library for our app? What would be our recourse then?
Our terms of service are listed at https://www.inboxsdk.com/terms. Specifically we will notify registered developers whenever we change our terms and you can use the old terms for some period of time.
In general though, we treat the SDK as its own product and have several competitors to Streak using it. We have millions of end users using apps built on the SDK from major companies (Dropbox, Hubspot, etc) so we're pretty careful with our terms.
The main goal is to retain the ability to maintain compatibility between different extensions being run by the same user and being able to respond quickly to changes in Gmail.
We use your InboxSDK and it is wonderful. Made development of an extension delightful, especially compared with the workflow around Outlook Add-ons and Wordpress Plugins.
By the way, there's a real opportunity to own the content side of gmail development. Articles on getting up and running, like building a basic Rapportive clone, would be incredible.
The library is remote loaded at runtime. We update it on our servers and the next time the user refreshes gmail, the library gets pulled down again.
EDIT: we have automated systems that help with this though making our turnaround time incredibly fast. We also see the changes to gmail before 99% of users so usually our updates are out before users even refresh to get the new version of gmail
That still doesn't make it reliable. If it's known the DOM changes regularly and breaks this api library, then why bother in the first place? Sure you can fix it yourself, but why not avoid the hassle and use something standard that doesn't break as much.
If it's acceptable that it breaks often then it's fine to use, but if you're expecting reliability, then no.
We were creating a chrome / firefox extensions to manipulate the gmail UI, the project was unrelated to email protocols (but I understand the snark ;-) )
This is a great library. Standardizes most of DOM manipulation/action triggering for everything Gmail does. Although Gmail's DOM changes sometimes like seibelj pointed out, you'll have the same issue if you roll your own implementation, so, for me, this is preferable.
If I remember right, it underwent a major rewrite a year or so ago, after which is has been very solid.
Creator of gmail.js here - thanks Nathan! I started working on this almost 3 years ago for a personal project. At that time there weren't many libraries available to work on top of gmail so I decided to compile all the helper functions into a standalone library hoping that after it goes on github, others might contribute.
After I did a Show HN in late 2013 a lot of people expressed interest in using this and ever since then, gmail.js has been used by a lot of individuals and startups. I'm really happy that people still frequently contribute to the library by adding features and/or reporting bugs and it has definitely gotten stronger and more stable over the past year (a very special thanks to Brent Kelly)
As a funny observation, the public sentiment towards gmail's DOM still remains the same as it was 3-4 years ago, and while I agree that it changes quite often, I do believe that the severity of those changes is heavily overstated. The disadvantage of an abstraction layer like gmail.js is that you always have to be on top of the DOM, but gmail.js is at a place where enough people are using it that if some major change occurs to the gmail UI, it'll be reported a lot faster because there is now a community using the library. In the past 3.5 years, only a single DOM change on gmail (about 2 weeks ago) has made a breaking change that affected all users from this library.
The whole thing is intentionally designed to be a single file where majority of the functions are standalone so if something goes wrong, anyone can easily understand how it works! That being said, a lot of the focus of this library is to abstract network events so developers making extensions on top of gmail can fire off events when a gmail user gets a new email, sends email, deletes etc. Those things rely very little on the DOM.
They have slightly different usecases. I used them both extensively and one is intended for adding a sidebar style chrome extension ala Streak, while Gmail.js is a simpler event based system.
I personally found both to be lacking in different areas, too bad they weren't merged together to cover a broader base.
Founder of Streak here (we make the InboxSDK). Would love to know which use cases we don't cover but do agree, both libraries were targeted at different layers.
Gmail.js lets you avoid doibng the DOM hacking yourself but is still fairly low level. The InboxSDK was targeted for people who are building apps inside of Gmail. We assume multiple extensions to be running and make sure they play nice, we handle auto updates to handle changes in Gmail and we try to guide developers to using standard Gmail UI styles. We have inbox support coming soon which will let you write your app once and have it work in Gmail and Inbox.
I had more clarity in my thoughts regarding this but it's been a year since I worked on the chrome extension and the startup folded. Thanks for your contributions to OSS.
Now for somebody to use this to write a Gmail extension that removes the hundred-levels-deep signature and reply chains that Outlook appends (and which evades Gmail's currently reply-collapsing).
I want to thank the author. Like many users, to repeat this project entirely on my own would be a nightmare. The GMail HTML source produced from the server is "minified". Random, meaningless class and id names. Author took on the burden and made Gmail more accessible for browser extension/add-on. I implemented a FireFox addon (for accessibility purpose for my senior thesis) using Gmail.js. Furthermore, the code is very readable and the API is extremely intuitive and well-designed.
Thank you!
https://www.sitepoint.com/mastering-your-inbox-with-gmail-ja... https://www.sitepoint.com/sending-emails-gmail-javascript-ap...