[PATCH 3 of 5] largefiles: use smaller files in test script

Carter, Eli Eli.Carter at tektronix.com
Tue Oct 25 09:12:58 CDT 2011


> -----Original Message-----
> From: gerg.ward at gmail.com [mailto:gerg.ward at gmail.com] On Behalf Of
> Greg Ward
> On Mon, Oct 24, 2011 at 11:13 AM, Carter, Eli <Eli.Carter at tektronix.com>
> wrote:
> > I disagree with these changes.  We need to test that the memory warning
> message is not displayed in the first case.  Your change of filesize eliminates
> that check.
> 
> Oops. So this means we need *one* "hg add" of a >= 10 MB file in a
> largefiles-enabled repo, right? But there's no point in adding a 23 MB file, or
> in adding a >= 10 MB file in a non-largefiles repo. So I may have trimmed too
> much, but I still think some of what I trimmed was a legit target.

At least one of the lfconvert tests would need to add a 10MB file in a non-largefiles repo for the conversion tests.  (I thought I saw a patch go by that moved the conversion tests to a separate file, but I haven't looked at it yet.)

> > Also, this changes the intent of the "Commit another file that should get
> automatically added as a largefile" test.  The point of that test was to test
> that a file is automatically added as a largefile _without_ any options on the
> commandline for largefiles.  That does mean a 10MB file is required to test
> that.
> 
> But the entire test runs with largefiles.minsize configured, so
> *nothing* tests the builtin 10 MB limit. I'm pretty sure I didn't change that --
> it was already like that when I started hacking on the test.

Ah, I hadn't noticed that.  Hrm.

> For the record, the intent of my patch was primarily to tweak test
> runtime/space requirements, and secondarily to test minsize < 1 MB.

I should have started my email with an acknowledgement that I understood that intent, and support the desire to reduce the test runtime/space requirements.  Just objected to the details. :)

But then, how do we want to deal with the need for a 2GB file testcase?  Right now, 2GiB-1B will work, but 2GB will fail.  Shouldn't we have a testcase for the 2GB-1B case?   Or is there a different place for the 'intensive tests'?

Eli



More information about the Mercurial-devel mailing list