More bug tracking etiquette

The title of a bug is important.  When you have a lot of bugs to manage it’s awkward to have to read the notes.  It’s kind of like email subjects.  Don’t you wish they actually gave you a decent hint of what the message was about?  Ideally the bug titles should give a good indication of what the problem relates to. 

Another thing to remember about the title is that you don’t need to overload it with information.  Bug trackers normally store quite a lot of fields about the bug in other fields.  Priority information and other classification information normally have separate fields available.  They can all be displayed on a bug list so you don’t need to clutter the title up with them.

Exactly what you put in the title depends on how you are classifying your bug reports in your system.  It’s worth discussing in your organisation so that you come up with a general set of guidelines.  If you have a good product/component classification scheme in use on your bugs then you only need to provide minimal information in the title about which system is affected.  For example a good bug report might go like this,

• Title: Crash when trying to add a 4th phone number to my contact list.
• Product: gmail
• Component: contacts list
• Reported version: beta
• Priority: High
• Severity: Trivial Bug
• Url: https://mail.google.com/contacts/

With that kind of information someone needing to manage the bugs can do so simply by looking at the title and whichever of the other fields are relevant.  A title of ‘Important: crash in gmail’ on the other hand would be pretty useless.

If the problem is serious remember to make sure that’s obvious.  I’m inclined to say if you have a crash make sure it’s mentioned in the title.  Actually that’s a bit misleading.  What I really see is that non-technical people don’t tend to distinguish between error screens and things not working as they expected.  If you see an error screen that should be clearly stated, that’s commonly the sign of what geeks would call a crash.  An error screen being something that gives a generic error message, or shows what appears to be lists of files and numbers and says an error occurred.  That’s as opposed to a specific error message saying something like ‘you need to type a number in the amount field’.  Error screens/crash reports are usually pretty easy to fix and an obvious error on the programmers part.  Programmers want to know about those first because they are a no brainer thing they need to fix.

One other minor point to note is that if you have discovered a serious problem like a crash while testing another bug report don’t tack it on as a side note.  Create a brand new case and clearly mark it as a crash.

NHibernate for Advantage Database Server

Note: this post has been in my drafts for so long I’ve changed company and so this isn’t strictly relevant to me now. I developed this in my own time while working at my previous company and I figure it may come in useful to someone.

I’ve just been playing about with NHibernate and Advantage Database Server (ADS).

I’ve done some initial code to make NHibernate work with both DBF and ADT files.

I’ve put a sample solution up with both DBF and ADT configuration files.  I’ve created a dialect and a driver for them.  I’ve also found it necessary to create an ID provider in order to deal with the DBF’s.  Since the NHibernate team recommends using a HiLo ID creator however it seems to make sense for the ADT dialect to also specify that by default.  I found it necessary to create the ID provider because ADS provides an odd way to lock records that isn’t done via SQL, it’s done via their data provider classes.

This dialect does the bare minimum to support the sample I found at the start of my NHibernate In Action book.  It does no more.  It doesn’t do anything like type mapping or any of the other things a dialect normally does.

I’m not sure what to do with this next.  I have tried running it against the NHibernate tests but there are so many failures because I’ve not implemented the basic things it’s a bit hard to see where to go.

For now I think I’ll see if I can use it in practice and then see where the holes are.

The one less than ideal thing that I’ve had to do is with the IIdentifierGenerator is basically copy the bits of code I wanted from the TableGenerator class and then modify them as required.  I get the feeling that having to lock records in the manner I’ve needed to isn’t particularly common though.

To see the project and example download it from here.  You also need to download the Advantage Data Provider from http://devzone.advantagedatabase.com/dz/content.aspx?key=20&Release=13&Product=4&Platform=11.

Web development tools

This is actually something I wrote in 2007 but it still appears to be relevant so I figured I’d kick it out there.  Since then the only major thing I can see that’s missing is the IE developer toolbar which is now a standard part of IE 8.

Fiddler

A web traffic monitor that hooks into the IE http stack and operates as a proxy server to allow other applications like Firefox to be monitored. Just seeing the traffic is really useful but it actually allows you to meddle with the data too! Windows only.

Firebug

A neat debugger for Firefox. You can analyze the DOM and set breakpoints on javascript.

Error console

The error reporting of your browser can often be useful. The error reporting on Firefox is especially useful.

Web developer toolbar

A useful Firefox extension that allows you to see all sorts of information about the website.

Selenium

Useful for automating testing. Allows testing in most browsers. If you copy over a blank Firefox profile before testing you can even reset the cookies and such like to make everything repeatable.

Ethereal

Actually this is useful for a lot of network problems. You can see what’s actually going over the wire. There is a version for most platforms.

Netcat

A kickass util developed by one of the guys from l0pht. So simple yet so useful, essentially just allows you to do all those socket operations from a command line util so you don’t need to create your own tiny programs. Does tcp/udp/connecting and serving. Buffers the line you are entering so it’s more useful than telnet.

Downloads for Windows seem to have dried up but you can always grab a port written in a scripting language like Python. The *nix downloads are still alive and kicking.

Advantage in .net for VO developers – using OLE DB driver

Note: this is an old post I’m clearing from my draft queue dating from 2007.  I’m not sure it’s really relevant any more but what the heck.  I think Advantage are on version 10 now.

If you aren’t using Advantage 8 and need to use Advantage 7 you can still use it all from .net. You just need to use the OLE DB driver that they provide.

A sample connection string for the OLE DB provider is,

Provider=Advantage OLE DB Provider;
Data Source=c:\datapath;
ServerType=ADS_REMOTE_SERVER|ADS_LOCAL_SERVER;
TableType=ADS_CDX;CharType=ADS_OEM;
FilterOptions=RESPECT_WHEN_COUNTING;
TrimTrailingSpaces=True

The main difference between this and the .net data providers connection string is the server type.

Ubuntu gateway

I’ve always known Linux was easy to setup as a router but until just now I’d never actually gotten round to figure out how to turn a box into a gateway. I knew how to setup the network side, setting up the routes etc. but the bit where you tell a box to forward packets from one network to another I just never got around to finding. It’s dead simple so I figured I’d better jot it down for the next time.

Edit the /etc/sysctl.conf and uncomment out the net.ipv4.ip_forward=1 line.

To set forwarding right now without a reboot or restart of anything you should then do this,

echo 1 > /proc/sys/net/ipv4/ip_forward

If you’re after a more detailed guide covering setting up your own gateway I found this information from http://technology.ezinemark.com/ubuntu-firewall-router-gateway-31d0966f5c3.html which covers the whole subject in more detail.