I can't offer any concrete advice on this, but you might want to check out MUSCLE, which does translations between it's own flattened muscle::Message class and BMessages. You might find some useful information there. e Marc Flerackers wrote: > Has anyone looked into the flattened BMessage format yet? I wrote a BMessage > implementation, but flattening and unflattening is still not compatible. > > This is what I found out for a message like the one posted at > //www.freelists.org/archives/openbeos/05-2002/fullthread209.html: > > Data Bytes Meaning > > 1(ascii) 1 version > BOFQ 4 flattened bmessage > ?? 7 ?? > DPUQ 4 what(B_QUERY_UPDATE) > ?? 5 ?? > > 0F 1 fixed chunk size? > GNOL 4 type > 04 1 chunk size > 06 1 name length > opcode 6 name > 1 4 data > > 0F 1 fixed chunk size? > GNOL 4 type > 04 1 chunk size > 06 1 name length > device 6 name > 4 4 data > > 0F 1 fixed chunk size? > GNLL 4 type > 08 1 chunk size > 09 1 name length > directory 9 name > 524992 8 data > > 0F 1 fixed chunk size? > GNLL 4 type > 08 1 chunk size > 04 1 name length > node 4 name > 524295 8 data > > 0B 1 variable chunk size? > RTSC 4 type > 08 1 chunk size > 04 1 name length > node 4 name > 08 1 chunk size? > 8 padded string? > > Some of the header information and the count for the entry_hdr structs is > still missing. Any ideas? > > Marc Flerackers (mflerackers@xxxxxxxxxx) > Software Engineer > ANDROME NV > > > >