La page se tourne.
Le prochain chapitre arrive…
Psst… appropriez-vous votre lecture.
Polices et thèmes se trouvent dans Apparence. Vos yeux ont aussi leur mot à dire.
Le prochain chapitre arrive…
Proto-danksharding. Added blob space so rollups can post data far cheaper than calldata.
Vérification de la lecture vocale du navigateur…
Cette lecture est actuellement disponible en anglais. L’interface utilise la langue choisie.
Lire l’original anglais →EIP-4844 introduced blob-carrying transactions, providing a dedicated data-availability resource alongside ordinary execution. Rollups can publish batch data through blobs instead of relying only on execution calldata. Blob data is committed to cryptographically, while the EVM accesses references rather than treating the entire blob as ordinary contract-readable calldata. The design separates a major rollup cost from general execution demand and establishes a foundation for later data-scaling improvements.
A rollup batch submission can consume execution gas for its transaction and blob gas for its data. These resources have distinct fee accounting. If many users compete for ordinary contract execution, that does not mechanically imply the same price movement in the blob market, although real workloads can interact. For an end user, the rollup may combine execution and data-publication costs into one quoted fee. Understanding that composition helps explain why different rollups or different batch conditions can produce different charges.
The blob mechanism requires network availability for a protocol-defined retention period; it is not a promise that every consensus node stores every historical blob forever. Rollup operators and archival services therefore need their own strategy for historical reconstruction beyond the required window. Nor does a low blob fee prove a rollup's execution or bridge is safe. Assess its proof system, data reconstruction process, settlement contracts, and administrative controls separately.
Protocol upgrades can also change resource parameters, so original launch limits should not be presented as timeless network capacities.