That would be great! Just to make sure -- if I will use -DG__NOMAKEINFO, will G__loadfile load my C++ sourcefiles, even if they have strange names (filename extensions) such as "/users/chris/tmp/scripAAAI.bpY_" ? (And will it even work if, by pure coincidence, the tempnam will be "scripAAAIx.dll" ?) Thanks, Christoph On Fri 2001-08-10, Masaharu Goto wrote: > Hello Chris, > > In that case, I'll implement a compile switch G__NOMAKEINFO > in next release. It will skip MAKEINFO access for most of the > cases. > > Thank you > Masaharu Goto > > > >Date: Wed, 8 Aug 2001 20:13:01 +0300 > >From: Christoph Bugel <chris@tti-telecom.com> > >To: Masaharu Goto <MXJ02154@nifty.ne.jp> > >Cc: roottalk@pcroot.cern.ch, rootdev@pcroot.cern.ch, meirr@tti.co.il > >Subject: Re: [ROOT] RE:CINT filename extensions > > > > > >You are correct, CINTSYSDIR was not defined. I do that on purpose, because > >I want to use only the required minimum, I don't want to need extra files > >at runtime if it is not strictly necessary. The only file I still need at > >runtime is bool.h, and I would like to fix that too, if possible. I just > >use the core of cint, (libcint), without the stl libraries, include files, > >without cint.exe. > > > >I think that in my situation the error message about not finding MAKEINFO > >occurs only when I call G__loadfile() with a filename containing an unknown > >extension. I don't use MAKEINFO even at compilation time, (I made a custom > >makefile), so I really want to avoid needing it file at runtime. I am > >trying to use a 'standalone' libcint, with as little as possible things > >needed at runtime. The smaller it is, the easier to understand :) > > > >So I am now trying to find a solution. I think I will first try to make a > >'my_tempnam' function that will guarantee a filename without a dot. > >Another option is to use the new G__load_text() API, which makes the > >tempname by itself. (What happens if cint generates a temp name with a dot > >inside, will it also need MAKEINFO?) > > > >I this doesn't work I will let you know :) > > > >Thanks, > >Christoph > > > > > >On Wed 2001-08-08, Masaharu Goto wrote: > >> Hello Christoph, > >> > >> The change you requested can be made easily. However, if it is > >> a matter of finding MAKEINFO file, you can simply define an > >> environment variable CINTSYSDIR. Cint looks for $CINTSYSDIR/MAKEINFO. > >> The error obviously looks like this setting is missing. > >> > >> set CINTSYSDIR=c:\cint for windows > >> or > >> export CINTSYSDIR=/home/xxx/yyy/cint for UNIX > >> > >> Thank you > >> Masaharu Goto > >> > >> > >> > >> >Date: Mon, 6 Aug 2001 14:17:13 +0300 > >> >From: Christoph Bugel <chris@tti-telecom.com> > >> >To: roottalk@pcroot.cern.ch, rootdev@pcroot.cern.ch > >> >Cc: meirr@tti.co.il > >> >Subject: CINT filename extensions > >> > > >> >Hi! > >> > > >> >Our system takes a C++ sourcefile, copies it to a tempfile (the > >> >filename is generated using tempnam()) and then ask cint to load the > >> >files with G__loadfile(). > >> > > >> >Usually this works fine, but once in a while cint will say "Error: > >> >cannot open /MAKEINFO". This happens only when the tempfile happens > >> >to contain a dot (.)! for example: "/users/chris/tmp/scripAAAI.bpY_". > >> >So I think this happened because cint tries to recognize the filename > >> >extension. (even thought it is just a random string) > >> > > >> >I know cint needs to look at the filename extension in order to know > >> >if the file is a sourcefile or a precompiled dll. As it is now, I > >> >don't need cint to load dlls at runtime. Would it be possible to tell > >> >cint (at compilation time) that *all* filenames (no matter what > >> >extension) are source files? > >> > > >> >Thanks! > >>
This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:50:57 MET