richardlehane / siegfried

signature-based file format identification
http://www.itforarchivists.com/siegfried
Apache License 2.0
224 stars 30 forks source link

misidentifications when running with -multi flag #117

Closed richardlehane closed 6 years ago

richardlehane commented 6 years ago

Testing on Greg Lepore's PRONOM FILES corpus have found that the -multi flag can cause misidentifications for a small number of OLE container files. These misidentifications don't reliably occur so looks to be some kind of race condition

richardlehane commented 6 years ago

Traced this down to the github.com/richardlehane/mscfb library. A global sectorSize variable made that package unsafe for concurrent usage.