EPG and XMLTV for FAST channels

What the guide is for

Two things, and they pull in different directions. For the viewer, the EPG is discovery: scrolling a grid and deciding whether to stop on you. For the platform, it is validation — proof that the channel has a real schedule running into the future rather than a loop somebody forgot about.

XMLTV is the common interchange format. It is a plain XML document listing channels and programmes, each with a start time, an end time, a title and usually a description, category and image. Unglamorous, and every platform ingests some flavour of it.

What belongs in it

• A channel definition with a stable identifier — changing it later is how a channel becomes a different channel as far as the platform is concerned.
• Programme entries with unambiguous start and stop times, timezone offsets included. Ambiguity here is the single most common defect.
• Titles written for a grid, not for an internal asset database. "Business News — Market Open" is a title. "MB_SEG_20260907_0731_v2" is a filename that reached production.
• Descriptions and categories, which drive search and recommendation surfaces on some platforms.
• Artwork where supported, at the platform's dimensions.

How far ahead it has to run

Further than most first-time operators expect. Platforms typically want days of guide data available at all times — commonly somewhere between three and fourteen, depending on who is asking — and they want it to stay populated rather than shrinking towards the present.

For a news channel this is genuinely awkward, because you do not know on Monday what Thursday's package will be. The normal answer is a structural schedule — recurring named strands at fixed times, populated with actual content closer to air — so the guide is truthful about the shape of the day without pretending to know the specifics.

The failures worth knowing about

• Gaps. A minute of unscheduled time between programmes; some ingests reject the whole file for one.
• Overlaps. Two programmes claiming the same minute, usually from a rounding error in a generator.
• Timezone drift. Times that are right locally and an hour out for the platform, appearing and disappearing twice a year at the daylight-saving boundary.
• A guide that stops. Data running out at the end of the week because generation was a manual task somebody stopped doing.
• Guide and stream disagreeing. The most damaging one: the grid says one thing and the channel plays another, which viewers read as a broken channel.

Where MediaBlaze fits

MediaBlaze generates the metadata alongside the media, from the same source of truth, so the entry describing a package and the package itself cannot disagree. For a news channel that is the whole point: the schedule is continuously repopulated as segments are produced and approved, rather than maintained by hand in a spreadsheet that is accurate right up until the week somebody is on holiday.