Posts from July 2007.

Switched to FogBugz On Demand / Workaround for changed scoutSubmit URL.

At the end of last week I created a brand new (free) FogBugz On Demand account, extracted my existing FogBugz MySQL database with mysqldump, and submitted it to Fog Creek to get it loaded up into my new account.

It took a few days before my new account was fully loaded, it seams Fog Creek have a little back log of uploads to do, obviously FogBugz On Demand has proven popular.

All is fine and dandy now, my support email is getting pulled into FogBugz On Demand directly from the POP3 account on my domain host (there was a little blip with the first couple of emails not being deleted from the POP3 server at first, but since then it’s worked flawlessly), and I can access my cases and get notifications with no problems.

However, there is one little gotcha, ScoutSubmit, the mechanism used by my applications (see RBScoutSubmit) for submitting crash data directly into FogBugz.

You see, my applications used to send their scoutSubmit data to my Mac OS X server sitting in my office at home, via a domain called support.imijsoft.com that I setup at my domain host to point back to my ADSL static IP address. This worked fine, occaisionally my ADSL would go down, so for an hour or two I might not get any crash data, but seeing as I can go weeks, if not months without a crash being reported, that wasn’t a big deal.

So, to make the switch seamless, I figured all I had to do was point support.imijsoft.com to the my new FogBugz On Demand domain via a CNAME alias, and I’d be home and dry.

How wrong could I be!

First, my old FogBugz server was at support.imijsoft.com/fogbugz/, and the new FogBugz On Demand server doesn’t have a /fogbugz/ path, oops!

Second, even once I had the new CNAME directive in place, when I accessed support.imijsoft.com FogBugz On Demand complained about there not being an account setup for support.imijsoft.com, oops again!

I contacted Fog Creek support to see if there was any way round the domain name not being a valid account issue, seems they are aware of it, and would love to be able to do something about it, and with a little work could probably get something in place, but there’s one big problem, SSL certificates. Currently, all FogBugz On Demand accounts are served up via HTTPS, as you would expect, and all are covered by the fogbugz.com certificate. Getting something in place that allowed other domain names to be used instead, with valid SSL certificates, that’s gonna be a nightmare, if it could be done.

The only option therefore would seem to be to change all my code to point to the new FogBugz On Demand URL, or switch back to using my own server.

I’m getting rid of my server soon (I’ll talk about that in a future post), so switching back is not an option.

I don’t want to change my code and release new versions of my applications to point to my new FogBugz On Demand account either, not only because I would then have to change and release new versions again if I decide FogBugz On Demand really isn’t for me (unlikely), but also because there are numerous versions of my applications already “out there” using the old support.imijsoft.com/fogbugz/scoutSubmit.php url.

So, the first (should that be second?) solution I came up with was to setup my support.imijsoft.com domain as a proper website on my domain host, and use a rewrite rule in a .htaccess file to proxy the requests through to FogBugz On Demand. This almost worked, I could get it working while accessing the website myself, but submitting data (using POST rather than GET) to scoutSubmit.php didn’t work, even with the rules taking into considderation the switch to ASP from PHP. I spent many hours last night on this, and then it struck me…

… I needed a proxy script, just like the one I’ve already developed for CaseDetective On Demand!

You see, CaseDetective On Demand is built with Flex, which runs in your web browser’s Flash engine plugin, which is very very serious about security. From Flash, you can’t grab data from any old domain, you have to have the authority to do so. To get data from a website without them putting a little file (crossdomain.xml) on their website that gives you permission, you have to use a proxy script. A proxy script is something non-Flash that you put on a site that you have control over and trust, which takes your request, forwards it onto the real destination url and then returns the results when they come back.

I built one of these little scripts for CaseDetective On Demand, so it wasn’t too hard to strip it down a little and make one for this job. You can download an example of the proxy script if you so wish.

The script works for both GET and POST methods of submitting data to scoutSubmit.php, I’ve placed this script in a fogbugz directory under my support.imijsoft.com root, with the root and fogbugz directory protected from prying eyes with a dummy index.html.

If you’re considdering switching from your own Linux or Mac OS X FogBugz install to the excellent FogBugz On Demand service, this script may just come in handy. You need to edit it before you use it to change the destination to your own FogBugz On Demand url, I think it’s pretty obvious as to where you do this.

Awsome new theme for this here blog.

Thank you N.Design Studio.

mnemonic

For over a year now I’ve been routinely typing the word mnemonic as part of my development work for a client.

Isn’t it ironic that I still stumble over this word when trying to remember how it’s spelt?

Also, wasn’t the Johnny Mnemonic film a complete let-down if you’d already read the story in Burning Chrome?

Just a thought.

In development: CaseDetective On Demand

I bet you didn’t see that coming! :)

I think it’s sometimes a good idea to conduct a thorough hard-hitting interview with yourself about a subject to fully explore said subject. There now follows an interview between you (U*) and me (I), where I pose and answer the questions I think you might ask of me, and I’m not going to hold anything back…

U: So, FogBugz On Demand is public, is there any likely-hood of a CaseDetective On Demand then?
I: Yep. CaseDetective On Demand is being developed as we speak, well not at this very moment as IMiJ Software’s sole developer is all tied up answering your inane questions … but it is being developed when I’ve got any time available.

U: Cool! Tell me more, what’s it developed in, I guess it’s not a REALbasic application then?
I: Your guess is right on the money, how clever of you! CaseDetective On Demand is being developed in Adobe Flex for that Rich Internet Application (RIA) experience. In fact, it’s not just CaseDetective On Demand that’s being developed in Adobe Flex, CaseDetective 2.0 is too.

U: Eh, CaseDetective 2.0 for FogBugz is being developed in Adobe Flex, how’s that going to work, I thought Flex ran in Flash?
I: Adobe Interactive Runtime, AIR for short. I can use pretty much the same code base for both CaseDetective 2.0 and CaseDetective On Demand.

U: Now that really is cool! If CaseDetective On Demand and CaseDetective 2.0 are being developed in the same technology, from pretty much the same code base, are they going to offer the same features?
I: Almost, but not quite. As it stands, I’m thinking of making CaseDetective On Demand a free taster of what’s in the installable CaseDetective 2.0.

U: So what’s going to be different then?
I: Nothing’s set in stone just yet, it’s very early days so things are very very likely to change as development progresses, but this is how I see things panning out…

CaseDetective On Demand will look almost identical, but obviously will be running in your web browser rather than on your desktop.

CaseDetective On Demand won’t be caching data like CaseDetective 2.0 probably will, it’ll grab “live” data from FogBugz when ever a selection changes. If it’s being run online, then you’ve got to have an internet connection anyway, right?

CaseDetective On Demand is unlikely to have all the report outputs of CaseDetective 2.0, these will be reserved for CaseDetective 2.0 as an incentive to download, but you will probably be able to get a good idea of what would be printed or extracted from CaseDetective 2.0, you just won’t be able to actually save it out.

U: Will CaseDetective 2.0 work with both an installed version of FogBugz and FogBugz On Demand?
I: Yep. In fact, if the installed version of FogBugz has it’s API URL available to be called from the internet, then CaseDetective On Demand will work with that too.

U: What if people find CaseDetective On Demand to their liking, and want it to become a fully featured application that they can access from any modern web browser?
I: If there is enough interest, then we’ll just have to see about making it so.

Plan A was to make CaseDetective On Demand a fully working online application, with a small subscription charge per month per FogBugz user id used (with the first one being free). But by plan X things had settled down to CaseDetective On Demand being a great way to get a taster of the full CaseDetective 2.0. Who’s to say that in the future plan Z doesn’t loop right back to something similar to plan A?

U: I’m sure you can call web services such as the FogBugz API from within a REALbasic developed application, why are you jumping from REALbasic to Adobe Flex/AIR for CaseDetective 2.0, surely development time would be less if you stuck with REALbasic?
I: You’re probably right, if I stuck with REALbasic as my development platform for CaseDetective 2.0 I’ll probably be finished a bit quicker than I expect to be with a complete re-write to Adobe Flex/AIR.

The thing is, if I stick with REALbasic I’ll probably not be able to offer CaseDetective On Demand, I simply don’t have the time to be able to develop a brand new application of this scale for the web, and make the extensive updates necessary to CaseDetective for FogBugz 6.0. By developing in Flex/AIR I’m killing two birds with one stone, as the rather ugly saying goes.

U: Ugh, yes that really is an ugly saying, thanks for that! I’m still not completely sold though, why not forget about CaseDetective On Demand and just update CaseDetective for the FogBugz 6.0 changes?
I: OK, you want the real reason? The huge improvements to FogBugz 6.0 scare the life out of me!

One of the most sort after updates to FogBugz for quite some time has been the beefing up of the filters, making them much more flexible, with many more options. The rumour is that FB6 will have some seriously improved filtering and searching capabilities, just the thought of having to develop the changes needed to duplicate this functionality with direct queries to the database scares me rotten. And there’s a lot more coming in FB6 than just filter and search improvements.

With that in mind, and the fact that the FogBugz has an API that can be called as a web service which includes such things as a list of filters, and retrieving case details for those filters, there is nothing for it but to switch to using the API. And if I want to support FogBugz On Demand, it is absolutely necessary as there is no access to the database.

U: But you’ve pretty much said all that already, surely that’s not the real reason?
I: Switching to the FogBugz API is a big job, and although REALbasic has good web service features, other technologies do it better, Adobe Flex being one of them. When I first heard about AIR (called Apollo at the time) my interest in Flex was piqued even more.

When I took a long hard look at Adobe Flex 2, I saw lots of stuff I really liked. It’s got a very nice IDE (Flex Builder, based on Eclipse), MXML is very flexible and quick to work with, ActionScript 3.0 is familiar without even having seen any before (it’s very ECMAScript compliant), and the optional chart components are fabulous.

It’s those chart components that finally tipped the balance. The charts functionality introduced in CaseDetective 1.1 has been a big success, but they are very basic, and frankly not as good looking or feature full as I would like. It would take me a humongous amount of work to improve the current charts, in fact they would most likely have to be re-written from scratch as at present they are a third party component that hasn’t seen much improvement of late. Flex’s charts are really good looking, easy to work with, and have many options for extension.

And besides, I’m still looking for a development platform that I can take forward and start consulting with when Informix 4GL finally dries up. Flex is much much more exciting to work with than Informix 4GL, and has a much bigger job market than REALbasic even at such an early stage in it’s own development, so I’m hoping I’ll be able to get some serious experience in Adobe Flex with this project and a couple of other projects I have in mind to show off to prospective clients or employers.

U: Ah-ha! Now the truth be told! You’re developing CaseDetective 2.0 and CaseDetective On Demand in the technology you’re looking to use for making a real living at for the next few years!
I: That is the truth m’lord. As much as I’d like to think CaseDetective could pay the the mortgage, it’s a long way off from it at the moment, I mean light years away at present.

U: What about backwards compatibility, will CaseDetective 2.0 work with FogBugz 4 and 5 as CaseDetective 1.3 does now.
I: Unlikely.

U: That sucks.
I: Yeah, I know. FB5 will be partially supported as there is a version of the FogBugz API available that can be downloaded and installed already, in fact the skeleton of CaseDetective On Demand has been developed so far using that existing FB5 API. But it’s very light weight, hardly any data is available.

From what I can tell, most FogBugz users upgrade within a few months of a major upgrade being available, so I can’t imagine there’ll be a lot of new customers wanting to use CaseDetective with FogBugz 4 or 5 in the not too distant future anyway.

I expect FogBugz 6.0 will be available quite a bit before I’ll be able to get CaseDetective 2.0 developed and out anyway, so I don’t think backwards compatibility is going to be a major issue, it might be the case that CaseDetective users upgrade their FogBugz install to v6 and then start contacting me about when CaseDetective will support it! I’m more worried about that.

U: Hmmm, that is tricky, how long of a delay are we looking at before you’ll get CaseDetective 2.0 out the door with support for FogBugz 6.0?
I: I really can’t say at the moment. Life is pretty darn busy, I’m in a full time contract for a few more weeks, with an extension being offered for a few more months, and I want to spend some time with my wife and baby daughter now and then, so development time is pretty scarce.

With some feature prioritisation, missing out the least used features for now, concentrating on the core of CaseDetective’s being, I might be able to get something out by the end of the year, but it’s more likely to be early next year.

U: Wow, that’s quite some time away until release, sorry to hear that.
I: Me too.

U: Is there any good news for existing customers?
I: Well yeah, of course!

I’m confident CaseDetective 2.0 will be a much nicer experience for the user, Flex/Flash/AIR has some usability features that can make an application much more enjoyable to use.

Those new charts should look great, and those much requested pie charts will finally be available.

CaseDetective 1.x will still be around, supporting FB4 and FB5 for a while. Although all development efforts are aimed at CaseDetective 2.0 just now, any show stopper bugs in CaseDetective 1.3 will be squished where necessary.

And CaseDetective 2.0 will be a free upgrade to existing users of CaseDetective 1.x.

U: CaseDetective 2.0 a free upgrade to existing users, that’s a nice gesture.
I: It’s the least I can do, considering the delay and what not.

U: Well, thanks for such a candid warts ‘n’ all interview, your time is very much appreciated.
I: You’re welcome, thanks for giving me the opportunity to talk to you, I enjoyed it.

Were my questions and answers simply not good enough, do you have some proper questions for me? Then please post your questions in the comments for this post.


* I was going to use Y, but I simply couldn’t cope with Y-I, “pet” kept popping into my mind.
Caching of data on the desktop may not make it into CaseDetective 2.0, it may have to come a little later. I want to make CaseDetective a lot lighter than it has been, this may take considerable work.
In fact, I enjoyed it so much I might just be doing this kind of thing at lot more often!

FogBugz On Demand

So, the cat is finally out of the bag, Joel Spolsky has announced FogBugz On Demand, a professionally hosted version of FogBugz.

In my opinion FogBugz On Demand is a wonderful idea, it vastly reduces the barriers to getting up and running with FogBugz, if you have an internet connection and a few dollars at hand ($21 per user per month), you’re all set. No install, no server needed, no backups or upgrades to worry about, sign-up and go.

Some of us have been privy to the existence of FogBugz On Demand for a few months now, as frequenters of the Business of Software discussion group were let in early to beta test it. The deal was very good for being a beta tester, two users for free, perfect for the MicroISVs and such like that make up the Business of Software group.

Although FogBugz On Demand is a great deal at $21 per user per month, considering you’re not needing to pay up front for licenses, don’t have to worry about maintaining a server, backing up the database, or upgrading software, I do think Fog Creek have missed a trick.

I think Fog Creek would do well to extend the free license deal to any new account beyond the existing 45 day free trial, maybe just a single license for free, but preferably two per account.

Consider the hapless developer who finds himself in a new project with no feature, bug or inquiry tracking facilities. Imaging that poor sole being able to create a FogBugz On Demand account in just a few seconds, and start entering features and bugs to be implemented or fixed, passing the URL to their customer(s), fellow developer(s) or tester(s). Happy days! Suddenly the developer can have a central repository of all the work they have to do, can keep track of discussions related to the cases and project design and development (via the in built discussion groups functionality), and know when they have completed the work for each release. Even if the client hasn’t got a system of their own to track the project effectively.

And who knows, after the developer has given the spare free login to their manager, development partner, prime customer representative or favourite tester, how many others on the team are going to want to get in on the act rather than just use the public interface functionality? Now the customer has given the authorisation for 5 users of FogBugz On Demand for the remainder of the project. Next project they add another 5 users, now those two free logins have turned into $210 per month, every month, for this one client of the developer alone.

And once they have got comfortable with the great features of FogBugz (and the fabulous additions coming in FB6), maybe they’ll decide to buy the self-hosted version for all departments in the company.

Just how many contractors are there floating around from poorly setup client to poorly setup client? I suspect metric butt-loads!

As I said, FogBugz On Demand is a fantastic idea, it just needs to get one foot in the door as a couple of free licenses (not just a 45 day trial), and I think it’ll do very well indeed.

***

UPDATE 2007-07-11 15:40: Both Joel and Eric have contacted me to let me know that there is a “Student and Startup Edition” of FogBugz On Demand, that gives you the exact same two users for free functionality that I was whinging about being missing. It’s an option within the “Your Account” page for everyone. It has not been advertised just yet for fear of overloading the servers at a time when lots of people will be testing out the new service.

So, all my millions of readers, don’t go sign up for your free two person FogBugz On Demand Student and Startup Edition account at the same time, or else you’ll get me into trouble!

***

How does FogBugz On Demand impact CaseDetective?

Those who know anything about CaseDetective for FogBugz will know that it currently talks directly to the FogBugz database. With FogBugz On Demand being hosted by Fog Creek, there isn’t going to be any way to get to those SQL Server databases, it’s just not going to happen, they’re locked away tight. In fact, Eric Nehrlich from Fog Creek was kind enough to email me as soon as the beta was announced to tell me that there fact, and offer to help me in any way he could in getting any requirements championed for improvements to the FogBugz API for FogBugz 6.0.

As such, I’ve been emailing back and forth with Eric and Michael Pryor for a few months as Michael made progress with the spec and then development of the upgraded FogBugz 6.0 API.

With any luck the new API for FogBugz 6.0 is going to have enough information for me to reproduce all the basic functionality in CaseDetective that my customers rely on, not only so that I can support the huge changes in FogBugz 6.0, but also FogBugz On Demand.

***

So, the fun begins.

Just how am I going to cope with such fundamental changes to the data retrieval method for CaseDetective?

Will the next version of CaseDetective for FogBugz use the same code base, or start from scratch?

Will the next version of CaseDetective for FogBugz still be developed using REALbasic even?

Will the next version of CaseDetective for FogBugz have any other improvements?

There’s only one way to find out … stay tuned for the next post!