The crew at Cortex, the authors of Clover, are pleased to announce the first beta release and public demo of their new product, FishEye.
- Posted by: Dion Almaer
- Posted on: March 25 2004 04:38 EST
FishEye delivers a unified view of your source repository that provides easy navigation, powerful search, historical reporting, configurable file annotation and diff views, change-set analysis, RSS feeds, and integration with your issue tracker.
FishEye is powered by Lucene and Jetty, and Cortex would like to thank the people behind these projects for their great work.
More info on FishEye at:
You can try FishEye out on our live public demo at:
- free? by Bruno Borges on March 25 2004 14:53 EST
- Statcvs by Julien Dubois on March 25 2004 17:26 EST
- Cortex announces FishEye, unified source control viewer by Dean Sheppard on March 25 2004 19:00 EST
- Cortex announces FishEye, unified source control viewer by ant on on March 31 2004 02:25 EST
- Cortex announces FishEye, unified source control viewer by Adrian Spinei on April 01 2004 09:21 EST
Is it free? opensource and etc... like ViewCVS? :)
Is it free? opensource and etc... like ViewCVS? :)FishEye will be a commercial product. But as is Cortex's policy, we will make licences for FishEye availble to OpenSource projects.
It's not free, it's alpha (according to the website)..... So that's yet another great advertisement on TSS...
Have a look at StatCVS instead (http://statcvs.sf.net), it's a sourceforge project which create a great website from a CVS repository. It's easy to integrate in a continuous integration build (it's an Ant task) - personnally I make it run every hour with CruiseControl.
Version 2.0 of StatCVS is still in dev, however I've been using it for a few weeks and haven't got a single problem.
StatCVS doesn't seem to allow you to actually browse the code so you'd have to use something else for that. Fisheye looks pretty attractive and has some cool features. I wouldn't pay a whole ton of money for it but I wouldn't deny myself or my team a useful tool just because it wasn't free.
StatCVS doesn't seem to allow you to actually browse the code so you'd have to use something else for that.It does let you browse the code. I just tried it on their demo.
There's statcvs (http://statcvs.sourceforge.net/) and there's statcvs-xml (http://statcvs-xml.berlios.de/manual/index.html). Although the projects seem related, it is unclear in what way. I think that statcvs-xml is a fork, but I might be wrong.
Unfortunately Statcvs seem dead. There is no activity there. There are a few problems related with it, but the most important is speed (mainly because statistics are not incremental, but one-time "shots"). I have installed statcvs to expose metrics for a few CVS modules at my current employer, ranging from 60KLOC to 400KLOC. Most of the projects began more than a year ago and the log files are huge. Statcvs simply hangs on some of the modules ("hangs" = not being able to generate the statistics in less than 6 hours).
So the longest statistic I am doing is on the last 6 months. Cannot go further with statcvs.
Statcvs does not offer zoom functionality on statistics. If you want to see only the metrics of the last week you have to run it again.
So my system currently computes metrics for : last 6 months, last week, last week, last 24h (4 parallel runs of statcvs for each module).
It effectively takes THE WHOLE FSKDNG NIGHT to build the metrics site. An incremental process using an indexation engine such as Lucene would be a relief for the poor server (it's running at load 3-4 during the whole night).
Eventually, it's going to cost us some money. I can perfectly understand. If the price is reasonable and the performance is ok, why not ?
Note that I do not want to bash Statcvs or something. It does its job for the price. But if a competitive product appears, hey, why not ? I'll let the management compute the damn ROI - for me it'll just a better product after all.
As far as I can see, StatCVS itself has no source code browsing. One of the demo just happens to have a link to ViewCVS. I am referring to the "sample outputs" at http://statcvs.sourceforge.net/#sample
CVS Monitor (http://ali.as/devel/cvsmonitor/) is probably a better open source alternative to FishEye than StatCVS, in terms of features. Sadly not written in java, but in perl
It's an interesting tool, however, it would have to be pretty much free for my consideration. It definitely has the "wow" factor, but it would lose its appeal fairly quickly.
Dean: I'd say that's a pretty close minded view. If a tool is useful, will allow a team to work more efficiently, help deliver projects quicker.. but costs money, you're saying that it is off limits?
I'm one of the first to praise open source, but I also realise there are a tonne of great (closed source) commercial products out there. I think you've got to put the cost of a tool in relation to the earning capacity it might open you up to.
I'm not saying that Fisheye is the greatest thing since sliced bread, but "it's gotta be free or else".
A prime example of this might be issue tracking, sure you could go with bugzilla, but I've yet to see an uglier issue tracking system. Jira however is a great product, it's got that extra polish that makes it a usable and powerful tool. A lot of time this is due to lack of UI polish that commercial products have (they simply hire a graphic designer to tidy things up).
Not all open source projects have ugly UIs. There are some open source products that have pretty decent UIs because they are scratching an itch the author(s) need in their workplace. DCL (http://dcl.sf.net) has a fairly good UI, as far as issue tracking software goes.
subversion support? comparisons to ViewCVS? scalability?
I am betatesting the animal so :
subversion - nope
viewcvs - viewcvs is to FishEye what a donkey is to a Lexus.
scalability - great because it works on incremental basis, all ASCII files and CVS logs are stored and Lucene-indexed. On the 0.1beta version with debug on, queries on a repository of >1.Mlines are almost instantaneous (cannot tell because it's not logged), summaries are recomputed in 55ms when small changes are made and indexes are reoptimized in 2162ms. A new LOC count graph is generated in 118ms (when it's not cached). Ok, it's a decent machine (2.4GHz Athlon w 1.5G RAM).
I suppose these figures will change once new features are added. But: it's a nice start...