Today the repository analyser uses the BlobContainer#readBlob methods to read data from the repository. The implementations of these methods are typically resilient to various failures such as a connection dropped mid-download, and will silently resume a failed download when it makes sense to do so. Such problems should be rare in practice; a repository that consistently fails mid-download will appear to work correctly but with reduced resilience and performance.
The repository analyser should avoid silently retrying on failures and should instead consider them as an analysis failure.
Today the repository analyser uses the
BlobContainer#readBlob
methods to read data from the repository. The implementations of these methods are typically resilient to various failures such as a connection dropped mid-download, and will silently resume a failed download when it makes sense to do so. Such problems should be rare in practice; a repository that consistently fails mid-download will appear to work correctly but with reduced resilience and performance.The repository analyser should avoid silently retrying on failures and should instead consider them as an analysis failure.