Change Of Mind Required
There must be a very serious discussion in the applied computer science sphere about this kind of issues. Security must be the topmost priority and "the latest hot craze" (such as running everything on a hugely complex(read: questionable security) JavaScript VM) must be questioned.
We must question the use of C and C++, as 70% of CVE exploits could be neutered by Memory Safe Languages.
Projects such as seL4 and CompCert should be considered the ideal to be emulated in things such as web servers, databases. Especially RDBMS servers have a history of being extremely insecure. Pushing out a new release, improving benchmarks was long considered imperative. Security was an afterthought.
The KISS principle should be employed wherever possible, as only simplicity makes certain security proofs possible.
Convoluted and highly insecure stuff such as OpenSSL should be questioned at the concept level. Do we really need these complex hairballs ?
Best practices such as formal scanners, parsers, regex checking should be written.