SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
CommodityQuant
Topic Author
Posts: 218
Joined: July 5th, 2007, 6:16 am

Coding reviews -- walkthroughs

February 21st, 2009, 1:56 pm

Among front office quants, how common is it for the code to be reviewed by a colleague, either in some structured walkthrough or in a more informal setting? I have worked as a quant for two organisations. In one, there appeared to be absolutely no reviewing of the actual code. If the trading desk was happy with the results, everything was fine. In the other, whenever someone finished a coding project, another colleague was assigned to review the code.I was quite surprised to come across a large organisation where there were no coding reviews. Is my surprise naive? Is it a common view that cross-checking of complex code is unnecessary? (Please don't ask me to give names.) I'm deliberately withholding mentioning the language involved because I don't want to give clues towards the identity of the organisation but the programming was relatively complex -- I'm not talking about simple VBA routines.CommodityQuant
 
User avatar
Cuchulainn
Posts: 62898
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Coding reviews -- walkthroughs

February 21st, 2009, 2:58 pm

Investing time in code review reduces maintenance costs and improves quality. I have no insights into how this works for others but in my own company's work up until now a process of continuous improvement is used.We strive to design good systems (and we visualise them in UML) and we oscillate between UML and code. The problem is: you must allocate resources to do the code review, and there is no time; Consequence is that new features are more expensive to implement. But this story is older than Coronation Street and East Enders put together... You can get major savings and be more productive if you have good products, of course. But... Easier said than done. The software industry is very young and its has to reach the level of let's say civil engineering. In 50 years time, maybe.
Last edited by Cuchulainn on February 20th, 2009, 11:00 pm, edited 1 time in total.
Step over the gap, not into it. Watch the space between platform and train.
http://www.datasimfinancial.com
http://www.datasim.nl
 
User avatar
gjlipman
Posts: 1102
Joined: May 20th, 2002, 9:13 pm

Coding reviews -- walkthroughs

February 22nd, 2009, 8:23 am

In my career I've seen disappointingly little code review - there is an assumption that if the output appears correct, then the code isn't too bad.
 
User avatar
bojan
Posts: 347
Joined: August 8th, 2008, 5:35 am

Coding reviews -- walkthroughs

February 23rd, 2009, 11:36 am

In the financial industry code review is very rare in my experience. (In fact so rare that I would be interested to hear which was the organisation/group that actually does use it!). In part this is due to the fact that often zero value is attached the source code. Instead development and maintenance of code is seen simply as an expense and all money spent on this is written off immediately. A cynic would say that some people quite like this situation because it allows them to become indispensable to the organisation they work for since only they know enough to modify/maintain their code. Obviously I'm not cynical at all and so would strongly disagree with this statement.
 
User avatar
RiskUser
Posts: 303
Joined: July 19th, 2006, 4:30 pm

Coding reviews -- walkthroughs

February 23rd, 2009, 12:47 pm

It depends on what you mean by reviewed by a colleague. If it is a FO<->FO review, then I would agree with your observation. However, a FO<->Model Validation Review can be quite a different beast. If you do an internet search on "Model Validation Policies" you will see what I mean.QuoteOriginally posted by: CommodityQuantAmong front office quants, how common is it for the code to be reviewed by a colleague, either in some structured walkthrough or in a more informal setting? I have worked as a quant for two organisations. In one, there appeared to be absolutely no reviewing of the actual code. If the trading desk was happy with the results, everything was fine. In the other, whenever someone finished a coding project, another colleague was assigned to review the code.I was quite surprised to come across a large organisation where there were no coding reviews. Is my surprise naive? Is it a common view that cross-checking of complex code is unnecessary? (Please don't ask me to give names.) I'm deliberately withholding mentioning the language involved because I don't want to give clues towards the identity of the organisation but the programming was relatively complex -- I'm not talking about simple VBA routines.CommodityQuant
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On