Bill Woodger wrote:
>For me, changing any compile option at the moment of going to Production
>invalidates all the testing up to that point.

Then along comes Java.... :-)

I strongly disagree with the word "all." I don't think that word in this
sentence is grounded in a reasonable, rational, informed assessment of
comparative risks and testing costs.

Consider also this important point: many businesses are moving much, much
faster than this rigid point of view would ever allow. Take a look at this
2011 video, for example:

The whole video is worth watching, but fast forward to about 10:00 for the
key statistics. According to the speaker, (the Web commerce
site, not all of AWS) deployed code changes into production on average once
every 11.6 seconds in May, 2011 (based on weekday deployments; they
evidently have a slower but still rapid deployment pace during weekends).
That was their pace half a decade ago. Are your current testing practices
and policies able to support that sort of business velocity or anything
vaguely similar? If not, why not? Are you helping your business compete? (I
believe at least a couple readers do work for businesses in competition
with Amazon.)

Amazon, the publicly traded company, has a market capitalization of $385.4
billion (as of October 17, 2016). Among companies traded on U.S. exchanges
it's currently #4 by that measure. True, its price-earnings ratio is over
200, i.e. the company isn't all that profitable. But that's yet another
problem if you're in competition with Amazon.

Timothy Sipples
IT Architect Executive, Industry Solutions, IBM z Systems, AP/GCG/MEA

For IBM-MAIN subscribe / signoff / archive access instructions,
send email to with the message: INFO IBM-MAIN

Reply via email to