Andreas Tille <andr...@an3as.eu> writes: > I think it was some enhancement to follow Aarons hint to use blastdb > format 5. However, the test keeps in failing for i386 and armhf > architecture. For armhf the log[1] says:
Ugh, it looks like format 5 is architecture-dependent -- attempting to point a 32-bit binary at a V5 database from a 64-bit binary fails with Error: mdb_env_open: MDB_INVALID: File is not an LMDB file even if I regenerate the database with BLASTDB_LMDB_MAP_SIZE set to a 32-bit-friendly value, as does attempting to point a 64-bit binary at a V5 database from a 32-bit binary. metastudent-data is far too large to make architecture-dependent if we have any other recourse, so it looks like we should explicitly use V4 here after all; as previously noted, this would entail replacing the formatdb command with makeblastdb -dbtype prot -in "$<" -out "$(@:.psq=)" I see a few other binary packages that contain BLAST databases and may need similar treatment: emboss-test (needs checking) hhsuite (architecture-dependent, so presumably OK) hmmer-examples (needs checking) kleborate (architecture-dependent, so presumably OK) ncbi-data (known safe, shipping pregenerated V4 databases) ncbi-rrna-data (ditto) openms-doc (needs checking) staden-common (needs checking) I'll take a closer look when I get a chance. -- Aaron M. Ucko, KB1CJC (amu at alum.mit.edu, ucko at debian.org) http://www.mit.edu/~amu/ | http://stuff.mit.edu/cgi/finger/?a...@monk.mit.edu