Re: CVS problems: inconsistent, bad host.

[John Huttley <john mwk co nz>]

Ronald de Man wrote:

> On Sun, Apr 18, 1999 at 12:24:22PM +1200, John Huttley wrote:
>
> > When compiles fail I tend to delete and re checkout.
>
> When compiles fail and you can't really see how to solve it yourself
> (for example by deleting .deps directories), then you should just wait
> a few hours or a day before you update your existing tree and try again.
>
> CVS versions are expected to break once in a while. That's normal.
> Trying to 'force' a correct build by deleting your tree and rechecking
> things out is a complete waste of bandwidth most of the time.

You might reasonably think so. But I have found this not to be the case.
It ssems that if the cvs gets too far ahead of my local version,  update is no
longer fully effective.
Possibly because obosolete  files are not correctly deleted.
The inconsistency in cvs, as a separate issue from broken code in cvs, may
aggravate the probelm.


> <Snip>

>
> As I read it, you were getting .23 for a while until finally .24 came
> through. This is perfectly understandable. The anoncvs is a mirror
> and is updated maybe once an hour, so it takes some time before it gets the
> latest changes. If it took much longer, maybe the system was overloaded
> by people madly updating CVS over and over again.
>
> > A totally avoidable load on the system.
>
> Yeah just wait a few hours before you try again.

I don't believe the '1 hour' bit for a moment. It may be _intended_ as '1
hour'.

This particular problem occurred half a day or more after the anouncement.

I claim that one or more of the mirrors is lagging by a substantial amount,
and that it is a fixable problem.

Regards
John