Hello,
A short few comments before I go into a meeting.
1. Yes, my problem was due to VMS!!! and the fact that Multinet has
security holes for spammers and the new version of multinet delivered one
mail message and crashed with no documentation other than something
obscure on the web. It sucked 2.5 days of my time which I am trying to
spend either on BRAT or learning the obscurities of linux.
On that note: your advice on using
nm <lib> | cat -n | grep <symbol>
then
nm -C libBrat_Base.so | cat -n | grep <number>
eventually led me to success.
The first thought that came to my mind was the very obscure vms command
lib/lis/names libBrat_Monitor.so
which was the vms command to see what was in the library. I was always
tempted to use something like
nm <lib> | cat -n | grep <symbol>
but vms always made me use those obscure commands. And I didn't have the
pleasure of splitting what I wanted into two commands that were equally
logical.
Enough of that
2. On the compiler, what you say might be correct. I cannot keep up with
3rd decimal changes. If the third decimal change makes it work, ok, But
my general conclusion is that if you mix compilers, you run into problems.
3. For Cygnus, I have successfully used that to compile brat under NT. I
hear whispers that it does not work as good on Win9x, but I don't have the
experience. The makefiles for Cygnus and linux are almost identical.
Got to go to my meeting.
Kris
Christian Holm Christensen wrote:
> Hi all of you,
>
> I've two things I want to convey to you. I also have a small comment
> for Kris, but it may be of general intrest, so read on if you will.
> Here goes:
>
> 1. BRAHMS Databases:
> ====================
>
> As you know, we'll do a dry-run on mid Febuary, and ASFAIK the
> intention is to have all software components in by then (at least in
> some alpha form). Hence I've been asked to deliver some code for the
> database interface.
>
> I'll try to have the code finsihed by the end of next week, and some
> write-up the following week. I'll commit my code to the BRAHMS main
> CVS in the BRAT directory for database handling ("db") and proberly
> bounce the tag to "VER-1-4-x".
>
> However, for this to work, we need to have one (or maybe more) MySQL
> server up and running, with some user set up with some priviliges. At
> the moment, I'm runninng a server on "hilux05.nbi.dk", and I have
> access as "cholm" with _all_ priviliges. This server is _not_ open to
> outside (that is, outside NBI) access, so don't try to use it.
>
> A _very_ prelimenary write-up is avaliable at
>
> http://www.nbi.dk/~cholm/brahms/bratdb.ps
>
> I think at least Konstantin and Flemming should have a look at
> it. Also, I'd like for Konstantin to provide me with a write-up (and
> perhaps code) on his Run database ASAP.
>
> More on this to follow :-)
>
> 2. NFS and AFS with Redhat Linux 6.1:
> =====================================
>
> Out SysOp at NBI (Bjorn Nilsson, email: nilsson@nbi.dk), is very
> interrested in any experience you might have on AFS and NFS on Redhat
> Linux 6.1. However, he's a burdended man, so if you're not sure what
> you may have is relevant, please email me first, and I'll may pass it
> along after a carefull deleberations :-o
>
> And a note for Kris:
> ====================
>
> Maybe you didn't get my mail the other day (due to your trouble in
> Texas - did it have something to do with VMS?), let me recab what I
> said.
>
> Previously, you've complained that you need to compile ROOT and BRAT
> with the same compiler for it to work. This is not enterily true. You
> need to compile with the same _major_ version of the compiler, but the
> minor is somewhat unimportant (except for compiler bugs
> ofcourse). Hence EGCS-2.91.66 (1.1.2) should work well enough with
> EGCS-2.91.54 (1.1.1), but neither will work with EGCS-2.90.29
> (1.0.3).
>
> And ofcourse not of these will work with GCC-2.92.2 (the re-united GNU
> Compiler Collection). The last point actually has do to with the glibc
> version, and the output from the compiler. GCC-2.95 makes some funny
> symbols in the compiled code. Doing a "nm -C <obj>" you're likely to
> see somthing like
> U printf@@GLIBC2
> U printf@@GLIBC2.1
> The same code compiled with any EGCS will show
> U printf
> Hence if you try to execute a program compiled with GCC-2.92, and you
> have glibc2.0 (and not glibc2.1), you'll get a SIGSEG (the infamous
> "Segmentation fault"). The lesson here: If you switch to GCC-2.92,
> switch to glibc2.1 also. On Redhat 6.1 EGCS-2.91.66 is the default,
> along with glibc2.1). This has _nothing_ to do with Linux, but is a
> common thing for all OS' (even Windoze and propberly also VMS).
>
> By the way, does any of you have any experience using Cygnus to
> compile ROOT and/or BRAT? If you do, I'll love to hear it (I'm trying
> to bring my own Windoze95 installationto it's knees - not very hard
> though).
>
> Cheers,
>
> Christian Holm Christensen
> ______________________________________________________________________
> Address: Phone: (+45) 35 35 96 91
> Sankt Hansgade 23, 1. th. Office: (+45) 353 25 307
> DK-2200 Copenhagen N Web: www.nbi.dk/~cholm
> Denmark Email: cholm@nbi.dk
>
This archive was generated by hypermail 2b29 : Thu Feb 03 2000 - 15:20:34 EST