BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.evolutio.pt//pycon-portugal-2026//talk//GU9KWG
BEGIN:VTIMEZONE
TZID:WET
BEGIN:STANDARD
DTSTART:20001029T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:WET
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T020000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:WEST
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-pycon-portugal-2026-GU9KWG@pretalx.evolutio.pt
DTSTART;TZID=WET:20260903T140000
DTEND;TZID=WET:20260903T143000
DESCRIPTION:Have you ever looked at a PySpark pipeline and assumed it would
  run top-to-bottom? You’d be surprised…it doesn’t. In reality\, Spar
 k builds and optimizes execution plans\, and this mismatch can lead to sub
 tle\, production-grade bugs.\nIn this talk\, I’ll walk through a real-wo
 rld example where reading and updating the same tables produced inconsiste
 nt results in production. We’ll see why this happens\, what Spark actual
 ly executes under the hood\, and why code that looks correct can behave un
 expectedly.\nJoin this session and you’ll learn practical design princip
 les to make pipelines predictable: introducing explicit materializations\,
  using declarative patterns\, or imposing physical separation between sour
 ces and targets.
DTSTAMP:20260904T104137Z
LOCATION:Auditorium
SUMMARY:The illusion of sequential execution in PySpark - Pier Paolo La Pas
 tina
URL:https://pretalx.evolutio.pt/pycon-portugal-2026/talk/GU9KWG/
END:VEVENT
END:VCALENDAR