4 ms·
You mean the support for MIME in stdlib, right? Based on the questions our customer support gets, quite a lot: just like we explained in the blog post. And base
by old-gregg 15y ago
You mean the support for MIME in stdlib, right? Based on the questions our customer support gets, quite a lot: just like we explained in the blog post. And based on the number of malformed and broken MIME we get via SMTP, even those that do aren't easy to understand for many.
Based on the calls/emails I personally had to answer, things like encoding, or making sure that HTML and text parts are specified in a certain order, or the aforementioned "Bcc mystery" bite quite a lot of people.
EDIT: forgot to add something: check out the "test mode" feature, you'd be amazed at how many people call us to flush up their email queue because they ran their unit tests on the production database! :)
- stephen_mcd 15y agoI'd love to see the actual question answered.
- old-gregg 15y agoJavaScript and PHP immediately come to mind. Ruby has a few MIME libs of a poor quality. Perl and Python have excellent MIME facilities but, as I said above, many developers simply aren't interested in learning MIME to begin with. We eliminated MIME on the receiving end earlier (see our "email routing" API) and now we've taken the next logical step: we have eliminated MIME on the other end. Besides, email is huge. In includes at least 4 non-trivial network protocols, several storage formats and a sea of proprietary ESP-specific extensions is error codes. There is no such thing as an "email support" in any language, only bits and pieces of uneven quality, therefore the most honest and shortest answer would be "all of them".
- jstedfast 15y agoIt's pretty trivial to bind MIME C-libraries like GMime for use in any language on any platform. http://spruce.sourceforge.net/gmime/ http://spruce.sourceforge.net/gmime/ It's a high-level library (unlike most other MIME libraries) which makes constructing and parsing MIME trivial.
- defen 15y agoYou're like the people who said Dropbox wouldn't succeed, because they had no need for it, because they could just setup rsync. Fact of the matter is that mailgun seems to have empirical proof that a lot of programmers have no clue what they're doing, and would be willing to pay for something that makes their lives easier. That's like lesson #0 of hacker news: people will pay for things that make their lives easier.