I built a bash script to manage deployments of multiple applications to a single Dokku server. I made this after I was paying like 30$+ a month on DigitalOcean, and wanted to move fast to a different VPS

What it does:

  - Import existing servers - Run ./deploy.sh --import ./apps --ssh your-server to clone all your apps, export their config (domains, ports, storage, postgres, letsencrypt), and env vars

  - Migrate between servers - Import from old server, change SSH config, deploy to new server

  - Multi-app orchestration - One config.json manages multiple independent repos

  - Hierarchical configuration - Parent settings cascade to deployments, children can override

  - Smart deploys - Skips unchanged apps by comparing git commits

  - Tag-based filtering - --tag staging or --tag api to deploy subsets

  - Secrets via .env files - Hierarchical: _api (shared) → api.example.com (specific)

Comments URL: https://news.ycombinator.com/item?id=46513346

Points: 4

# Comments: 2

Weaknesses / Counters:

1) Surrogate endpoint only — HPV PCR positivity is not a clinical outcome; no CIN2/3, no cancer, no mortality measured

2) Correlation ≠ causation — HPV-cancer link is epidemiological association; Koch's postulates not fulfilled in traditional sense; detecting DNA doesn't prove pathogenic activity

3) PCR detection ≠ disease — Transient HPV infections are common and clear spontaneously; most HPV-positive women never develop lesions or cancer

4) Type replacement signal ignored — 66% higher incidence of non-vaccine HR types in vaccinated group is dismissed rather than investigated as potential clinical concern

5) No long-term clinical follow-up — Cervical cancer takes 15-30 years to develop; this 7-year study cannot assess actual cancer prevention

6) Confounding in vaxxed vs unvaxxed comparison — Unvaccinated group is small (n=859), likely differs in health behaviors, screening adherence, socioeconomic factors

7) Circular reasoning — Vaccine "works" because it reduces detection of the types it targets; says nothing about whether those types were actually causing disease in this population

8) Assumes HPV16/18 reduction = cancer reduction — Untested assumption; clinical benefit must be demonstrated, not inferred from PCR

9) High baseline HR-HPV in vaccinated group unexplained — 32% prevalence of other HR types suggests substantial ongoing oncogenic exposure despite vaccination

10) Genome validity unestablished — HPV reference genomes are in-silico constructs assembled computationally; never validated by sequencing purified, isolated viral particles; PCR/sequencing performed on mixed clinical samples where true origin of amplified fragments is indeterminate

Comments URL: https://news.ycombinator.com/item?id=33618211

Points: 2

# Comments: 0

The mortality is as bad as in 2005 roughly!

Comments URL: https://news.ycombinator.com/item?id=26576764

Points: 47

# Comments: 64