ra-test.exe deadlock condition

2017-01-31 Thread Stefan
Hi, I've been looking at the cause of a deadlock when running ra-test.exe with -fs-type=fsx (trunk version). The most important findings are summed up here atm [1]. The issue was discussed with brane and danielsh on IRC (thanks for your time, once again). As far as my current understanding of t

Re: verify_as_revision_before_current_plus_plus() on a production repo?

2017-01-31 Thread Julian Foad
Daniel, thanks for your comments. Daniel Shahaf wrote: Julian Foad wrote on Mon, Jan 30, 2017 at 21:40:00 +: it is a little "tricky": in particular, it opens a second FS instance and fakes the "youngest revision seen" field and then relies on this FS instance never reading the "current" fil

Re: verify_as_revision_before_current_plus_plus() on a production repo?

2017-01-31 Thread Daniel Shahaf
Julian Foad wrote on Tue, Jan 31, 2017 at 10:22:06 +: > Daniel, thanks for your comments. > > Daniel Shahaf wrote: > >Julian Foad wrote on Mon, Jan 30, 2017 at 21:40:00 +: > >>it is a little "tricky": in particular, it opens a second FS instance and > >>fakes the "youngest revision seen" f

Re: verify_as_revision_before_current_plus_plus() on a production repo?

2017-01-31 Thread Julian Foad
Daniel Shahaf wrote: I agree with your last sentence, but I think you misunderstood me. I was simply arguing that the "root->rev is younger than value in 'current' on disk" is a situation that can happen in normal API usage, and therefore calling svn_fs_verify_root() in that situation is *not* un

[PATCH] add support for zlibstat.lib in install-layout for Windows

2017-01-31 Thread Stefan
Hi, the zlib project's visual studio file creates static libraries named zlibstat.lib by default. Atm SVN supports this, if it determines zlib is in source-layout, but doesn't support it, if it determines files are in install-layout (the detection being based on whether the zlib-path contains an i