Andreas Pakulat wrote:
> On 21.05.07 11:39:38, Matthew Woehlke wrote:
>> Done. I think we can also now remove RST::Valid and
isValid()?
>
> Yes I think so.
Done.
>> Maybe we should instead take both limits at once,
with default
>> parameters for each (make the revision limit a
pointer I guess, default
>> = NULL)?
>
> No, also the VcsRevision is not a limit, its more like
"from" and "to".
> And having them in 2 separate methods makes it more
clear that they mean
> different things. One is for fetching the last N items
in the history
> starting at revision, the other one fetches all history
items between
> two revisions.
My concern is just that if the VCS doesn't support 'from,to'
in log(),
you are stuck waiting on it to fetch 'everything'. Hmm... on
second
thought, any objection to requiring it to be kind-of like
showLog()? Use
a reasonable limit internally if the VCS does not support
'from,to' and
keep making limited calls until the 'to' revision is
reached, then truncate.
>> log() should be ok now (see above). Is diff() ok
as-is, or am I missing
>> a use case?
>
> No, I think diff is ok too. We might need to document
that dstPath
> doesn't need to be given, if 2 revisions are given. Or
do we force users
> to provide the same path for both (path == repo or
local path).
I was assuming you always provided all parameters, even if
that means
passing the same arg twice. I guess we could also document
that empty
dstPath means 'use srcPath'.
--
Matthew
"Nobody expects the traditional Bourne shell!"
_______________________________________________
KDevelop-devel mailing list
KDevelop-devel kdevelop.org
https://barney.cs.uni-potsdam.de/mailman/listinf
o/kdevelop-devel
|
