back again (again)

Well I'm back again, the little app I was working on had stopped being fun and as I was doing it in my spare time I decided to stop...

However I recently had to choose if I was going to renew this domain and I was doing a couple more fun techy things so I thought I'd re-start the blog and use the impetus of having shelled out for a .best domain for another year as a way to return me to the blog and to doing tech stuff in my spare time.

Not entirely sure exactly where I'm going to go with this. I will probably try and get the app up and working again, although it is successfully (ish) running in its hosted environment (lunch.ben.best) the version locally isn't working and upgrading my ageing mac appears to have killed off Git. So my plan for this evening is to get things up and running again. I've got an idea for a different app and I want to have a play with some other languages (I've been meaning to look at clojure and scala serverside and coffeescript for ages).

So I'm going to spend this evening getting things up and running again, I've just installed a new ssd into my ageing laptop and that is feeling a bit like a new machine. I'll let you know how it goes and if what I decide to do.

In the meantime here's a short, silent video of me replacing the hard drive:

broke it again...

Continuing with the let's see how bad I can mess it up plan that I've been working with so far, this evening I've fixed one thing and really broken lots of others.


I now have an app that although it correctly describes things so has got rid of the depends on rubbish and replaced it with done before which is what I needed it to be. Unfortunately it won't now let you save a new step (and I don't like that description either).

In positive news I can continue playing tomorrow so I'll hopefully at least get something saving.

I did this evening manage to get some things right. I  managed to get rid of a load of the errors that I was having to ignore in the code.  These weren't actual errors, they were jslint (a tool for telling you where you've written your code badly) not being told to correctly ignore some errors.  I'd previously tried to fix this and failed, actually getting this fixed meant I could see the proverbial wood for the trees and pointed out some actual problems that I'd been missing because I'd ignored all the errors.

The other thing that this has pointed out is that I really (really really) need to add in some unit testing.  This is starting to get embarrassing, I don't actually know exactly when I broke saving things, I think it was this evening, I can role back code changes and find out but that's a faff. Proper unit tests would have highlighted this to me as soon as I broke it.  Fortunately I pointed this out about 3 months ago, less fortunately I paid no attention.

As a sort of PS you might have noticed that I turned ads off.  The reason for this was that I no longer needed them. I'd played with them, figured out how to get them working and better than that realised how to set them up and how to turn them off.

I've checked in the current (broken code base) here:
https://github.com/benjimouse/lunch

managing a mailing list

This was specific advice I wrote to a friend who is starting to look into email marketing.

As with all things it sort of depends exactly what you're trying to do. I'm going to imagine that you're trying to send out a weekly email to around 3000 people.

What not to do:

Don't try putting their emails into the bcc field.

It's really easy to mess this up and put it into the cc, meaning that you've sent everyone each others email addresses.

Also doing it that way you end up having to deal with unsubscribes, people changing email address etc. all yourself.

What I'd suggest:

Look at mailchimp.com and campaignmonitor.com

They're both good, I've used both in the past, I have a slight preference for mailchimp as their ceo bought me pudding once but look at both and decide which you prefer.

Use one of their included templates it's free or nearly free, it means that you know it works with their site and has been tested to look good in most mail clients.

After you've sent out a few you might decide that you want to create or have created a specific template.  That's great and I'm happy to help with this (by which I mean I can recommend some people), but I'd wait and see if you're happy with what you're doing first and then look at it.

Test, test, test.

Create a small group (3-4 people max) of friendly people who can phrase criticism well (I'm happy to be on this) . Before you send any email out to a large group send it to these people and then email them as a group (this time cc them in) asking for comments.  This should mean that some mistakes get caught (some will always get through).

Allow people to unsubscribe easily

This should be covered by the template, the reason you want this is that if you don't do this then people will mark your emails as spam.  That means the big providers will start to see all you emails as spam. As well as this mailchimp, campaign monitor or whoever will stop providing services to you (they'll warn you first) they can't afford emails they send to be marked as spam.

Make your emails relevant

Following on from the above, there is no point in sending an email to a group of people that have never been to your venue a discount for those returning for the 25th time. (There are better examples but this was the one that came to mind).

Both Campaign monitor and mailchimp allow you to split your subscribers into groups, do this. It will take time and be frustrating but it's worth doing.  It will allow you to send more targeted emails that are more relevant and therefore produce (hopefully) better results.  Mail chip also definitely allows you to add code to add subscribers directly from your website, you should do this as well and if you can have them put into specific groups.

Look at the stats

After you've started sending the emails it's important to look at the stats that are generated.  It will allow you to start refining what to put in and when you send emails.

That's probably more than enough for a starter and no doubt contains some irrelevant bits that you already know and others that you don't need to.

playing with meteor some more...

Continuing with the playing with meteor.
I've now deployed where I am to http://benslunch.meteor.com/ I had a bit of a problem with this as when you deploy to meteor you need you username.  When you login on the website you use your email address, however when deploying you need your "username" something I set up about a year ago and haven't used much since. I couldn't remember or find it on the meteor.com website.
Eventually as is often the case stack overflow showed me the way.
It's slow going in between cinema trips (go and see Paddington, it's very good) and work, but I'm getting there and I'm really enjoying it.  I continued with the clean code book last night (there's that pay me money link again) it's interesting but I don't think I'm quite getting it yet.  I've also kept the github repo up to date if anyone is interested.
I've ended up creating a helper function in javascript (in the step_item.js file) that I'm really not happy with.  It works and so I'll keep it there for now with an eye to refactoring later.  There's a couple of ways I am currently thinking of doing that: Putting it into a global client scope or putting the logic into the template. I'm not over the moon about either of those and I'm hoping as I continue with the tutorial book (discovermeteor.com) I'll find a better solution.

I know that meteor does allow me to run unit tests, but I've not got to that point in the book.  If I'm still getting frustrated with the code clean book I'll start to skip ahead.  At the moment I'm doing this project for fun so I'm doing the fun bits, I'm a little concerned that I'm going to spend a fair chunk of time writing unit tests for existing code, something that I have depressing experience of and am not really looking forward to.