On Wednesday 26 January 2011 11:27:28 Leighton Pritchard wrote: > Just in case anyone else runs into similar problems (or there's an > opportunity to fix the underlying problem), I thought it might be useful to Phew ... thanks Leighton for that detailed summary. I'll go through and see whether I can ease a few things (if I find out out, that is). > Starting from a pretty standard CentOS5 with nearly all CentOS base/EL > packages (we're discouraged from straying by IT policy), flex and BOOST > versions are too old for MIRA's liking. Too old? Oh dear, what do they install in CentOS? > Both are straightforward-ish to > install from source, but the standard flex installation does not place a > symlink for flex++ (which should point to the flex binary), so the > configure script reports a version mismatch until this is fixed. Argh, the old flex vs flex++ nemesis. If only I know how to correctly check for this kind of interesting installations. Fortunately that does not happen all too often. > Having built BOOST with > [... snip boost problem ...] > I looked into the configure script, assuming that there must be some issue > with recognition of the libraries, but rapidly decided that messing around > in there was a bad idea. Yes ... I am relying on the autoconf BOOST macros and they are also really, really difficult to understand. When they work, everything's perfect. But when they don't, wrapping one's head around what they might be doing and why they fail ... well, I've given up. BTW, the amount of debugging output related to BOOST in the configure.in script gives a hint on the amount of work I already put into it :-/ > I hope this is useful to someone. This mailing list also exists for exactly that reason: allowing people to find it via Google when they have a problem ... and find a solution. Thank you for the detailed account ... at least for flex++ I think I might have a silution. > If anyone reading this spots something > simple I've overlooked, I'd be grateful for the pointers. [X] Me too! B. -- You have received this mail because you are subscribed to the mira_talk mailing list. For information on how to subscribe or unsubscribe, please visit http://www.chevreux.org/mira_mailinglists.html