We're in the process of moving to a new issue management system at work. Our old Access-based one was written in-house and is quite... clunky. (I'm not supposed to say that because I'll step on someone's toes. But if I write piece-of-shit software I'd rather someone tell me so I can fix it rather than pretend everything is ok.)
Anyway, the new system is Rational ClearQuest.[1] We've gone through some sort of process to settle on this software with a committee and meetings and such. (I'm sure information about it is on the internal wiki but I'm not interested.) And our project is the pilot for it so all our new issues will be entered into ClearQuest and hopefully our feedback will help shape the future direction of the product.
As the pilot team we needed a demo and overview of the system and late last week we got it. But keep in mind that in the best blogging tradition I haven't actually used the system yet. My opinions here are based on notes I took during the demo and conversations we had during and after it.
1. The standard interface is a Win32 client rather than a web browser. (This is almost a deal-breaker for me.) Plus, it looks like ass -- probably why they don't have many screenshots on the website...
Apparently there's some sort of web portal but we didn't see that. According to the datasheet CQ web access works with IE, Mozilla and Safari (!), so it might be okay but given the state of the Win32 client I'm extremely skeptical. I'm also unsure how much pain the IT department is willing to bear to deploy it, particularly if it depends on WebSphere...
2. Searching is case-sensitive. So if you look for a project 'mercury' you won't find 'Mercury'. Jesus, how can someone design something like that in this day and age? (Another potential deal-breaker.)
3. There's no 'search' box. Instead you have to create a query (using another awful interface) or reuse queries that you or someone else have stored for later use. And there's no way to just jump to issue number N without using queries. Again, another straw on the deal-breaker camel's back.
4. The client doesn't seem to use standard Windows GUI controls. So you can't type in dropdown boxes and quickly navigate the list of items. Since a number of the dropdowns have tons of possible values navigation slows to a crawl. I hate Windows applications anyway, so this just makes it even more painful.
5. File attachments seem to be stored in the database rather than a filesystem. And given typical BLOB issues we were given a general "don't use too many attachments" warning. Yeah, it's not like people need that or anything...
Much to the developers' surprise though, the Win32 client sent the file to the right application when we asked it to 'Open' one of them.
6. There are a ton of labels that are obviously just slightly-cleaned variable names, if not direct variables. (This may be a customization issue.) So you'll see things like 'Is_Issue_Foo_or_Bar' on a form or a directive to change to the 'To_investigation' state.
7. I don't think you can create a printable version of an issue or a query without Crystal Reports installed. But I could be wrong about that.
8. There's no screen with all information about an issue in one place. Just a dialog with about 10 tabs. While I guess tabs have been used to good effect somewhere they're generally just terrible, especially when all the tabs deal with different data attached to a single record.
9. There don't seem to be any shortcuts for commonly used data. For instance, in a number of places you're asked to reference a product, and the product includes version numbers so you'll have 'Mercury 1.00', 'Mercury 1.01', 'Mercury 1.02' as separate entries. Since many developers are working on the next version of the software it would make a lot of sense to have a 'Mercury LATEST' so you never even have to think about it.
10. The client exposes you to data and relationships you don't need to see. This is a gigantic usability problem and IMO the problem with many user interfaces because it makes the common case really hard. (All software should subscribe to: "Make easy things easy and hard things possible.") There are so many ways this manifested itself but I only jotted down a couple.
Here's one example: when entering an issue you have to create the issue (title, description plus some additional data) and then associate it with one or more "change requests". Your issue might touch several different products (or versions) and each one of these gets a change request. Makes sense from a data management standpoint, no problem. But the steps I have to take to create an issue are (from memory):
Hey, I've got a great idea: I want people to enter solid issue reports so I can fix and improve my software the best way possible. So I'll create a system which drives them to absolute rage when they enter issues!
More examples on this, just with identfiers:
Finally, one thing to keep in mind is that ClearQuest seems more like a toolkit than a finished system and many of these problems might be fixed with further customization. If they are I'd be happy to post about this again.
[1] Sorry Atlassian folks: I would have recommended JIRA in a heartbeat, but I had no say in this. One of the hazards of working for a larger company is that you'll probably have less of a say in the company-wide tools you use.
Posted from cwinters.com; read original
In my time I have used
The only issue I have with bugzilla is that it is targetted at non-perl users and rather than use good perl practice (CPAN modules, etc) it is all a bit hodge-podge internally even if a well documented podge.
I'd really like to get the chance to rewrite it in a couple of Class::DBI modules, using stuff like Email::Send so that it doesn't have to do quirky voodoo via the command line to send and receive mail and can generally use the good stuff on CPAN.
I'd also just like to have an API to access the bugs and users through in case I needed to integrate it with other perl applications or write more extensions.
Re:hmmm bugzilla
lachoy on 2005-01-17T20:21:06
> Why would anybody want to use a product like that?Personally I think it's a combination of, "It's from IBM so it must be good", "It costs a lot of money so it must be good", and the idea that simple tools won't be able to "keep up" with the business needs. I'd say this is mostly managers, but I do know another developer who in a previous job has used ClearQuest (along with ClearCase, the Rational version control system) and likes both of them quite a bit.
