Re: patch: global -u on win32 with src root at drive root

[Kai Tetzlaff]

Hi Shigio

> Hi Kai,
>> so far i was just looking at an easy fix for the problem at hand. But
>> yes, maybe it's worth it having a closer look. So i started doing this
>> and actually found that on unix, there's also a problem with having
>> GTAGS at /GTAGS. setupdbpath in getdbpath.c bails out with GTAGS not
>> found (line 277).
> 
> I also confirmed it. It seems that setupdbpath() does not look for GTAGS
> in the system's root directory. In a word, it does not admit tag files
> to be put in the system's root directory.
>
>> Now, on unix it might be questionable to have a GTAGS db at the root of
>> the file system.
> 
> Window version should also do the same behavior as UNIX version.
> That is, it should die with the following message:
> 
>         global: GTAGS not found.
Yes, that's what my first patch was supposed to fix (well, it actually
does).

> 
>>                  But on the other hand there's nothing which really
>> forbids this setup. So, if you think that this should be fixed too, i
>> guess i could have a closer look. Even though the fix which will result
>> out of this will probably have to change some more parts of the
>> root/dbpath detection code.
> 
> Do you need to put tag files on the system's root directory?
I don't need it for unix based systems (actually i'm not working on unix -
but
even if i would - there's no need to put sources at '/'). But under
windows,
the typical setup here at work is that a drive letter gets mapped to the
source
tree. We're using clearcase here, and on Windows this is the default
behaviour.
Of course there are ways around this, but i'm not sure i can convince
myself (and
co-workers) to switch to a different setup. So i would rather like to get a
fix
implemented in global (even if it finally means to maintain a set of
private
patches).

I played with this a bit this morning. I think i already have a working
version
which changes the behaviour of setupdbpath so that all paths it generates
(cwd,
root, dbpath) end with a '/'. It requires some additional changes to the
current
code. But then it solves both problems (windows and unix) with GTAGS at the
fs
root and actually does not require any platform dependent switches (so
far). But
it needs some more analysis and testing ... I can provide this to you if
you like
to have a look at it. Otherwise i would just play around it for a couple of
days
before sending it to the list.

> 
> --
> Shigio YAMAGUCHI <address@hidden>
> PGP fingerprint: D1CB 0B89 B346 4AB6 5663  C4B6 3CA5 BBB3 57BE DDA3
>