-
-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
In light of the Moq issue yesterday[0] I'm interested to understand why the consensus seems to be so in favor of keeping packages up-to-date in software.
The common explanation I see is it "keeps you up to date with security and bug fixes".
But in practice this seems to just involve most orgs mandating Dependabot and mindlessly updating every dependency when a new version becomes available. (Yes in an ideal world you code review every change in every dependency, but... I mean, let's be real here. Just take the update frequency of the AWS SDK packages in isolation, very few orgs are actually doing this)
As a maintainer of an open source library I know most releases are a crapshoot, they're just as likely to contain new bugs and flaws as they are to fix old ones.
So staying up-to-date seems to open up codebases to far greater risks than outdated dependencies:
1) Zero days, a new package launches with some critical security flaw that isn't going to get noticed for some time.
2) Supply chain attacks, old packages are generally immutable. Therefore most supply chains attacks seem to involve take-overs of existing package (name)s by disgruntled or new hostile 'maintainers'. The new versions are far more at risk.
3) New bugs, the dirty truth of OSS is most work is done by unpaid people with little time or ability to focus. Most software isn't formally verified. New updates are a risk.
In addition the old version is a known quantity. Unless you know absolutely the version you are running is compromised (log4j, OpenSSL) what benefits does updating actually bring? The default presumption that version number goes up is better seems like yet more security/compliance cargo cult behavior.
What am I missing here?
[0] https://github.com/moq/moq/issues/1374