Kris Zyp wrote:

Yes, I will work on upgrading this patch to 1.0. =20 However, I believe in order to realize optimal Windows performance with=

out regression/slow-downs, I will also need to disable the use of NTDLL (= and use direct

CreateFileMapping memory maps instead), as mentioned=C2=A0 http://www.o=

penldap.org/lists/openldap-bugs/201902/msg00009.html. You had mentioned i= n that thread that

the NTDLL code wasn't destined for release, is that still the case (it =

is still in the mdb.master branch)? If you are intending to release (1.0)= with the

NTDLL-based memory maps (which is understandable, can certainly be an e=

asier path for Windows users getting started), I will also need to add th= e compiler

option to disable it. Anyway, I will work on creating a PR with both th=

ese patches.

Might be interesting to investigate why the behavior with NTDLL is so muc= h slower on Azure. Since the Win32 APIs are implemented on top of the NT APIs, you would expect th= em to perform identically. Perhaps there's some additional flag that the Win32 API is passing to NT = that is needed on Azure.

=20 Thanks, Kris =20 On Mon, Feb 17, 2020 at 8:14 AM Howard Chu <hyc@symas.com <mailto:hyc@s=

ymas.com>> wrote:

=20 Kris Zyp wrote: > Sorry to keep pestering, but just pinging about this patch again,=

as I still think this fix could benefit windows users. And at this point= , I think I can

say we
> have tested it pretty well, running on our servers for almost a y=

ear :).

=20 Looks like this patch is against the 0.9 release branch. I hit a bu=

nch of conflicts

trying to apply it to mdb.master. We'll be stopping work on 0.9 soo=

n, and getting

LMDB 1.0 out the door finally, so can you please verify that your c=

hanges will work

on mdb.master as well?

=20 > Thanks, > Kris > > On Wed, Sep 18, 2019 at 12:56 PM Kris Zyp <kriszyp@gmail.com <mai=

lto:kriszyp@gmail.com> <mailto:kriszyp@gmail.com mailto:kriszyp@gmail.co= m>> wrote:

>
>=C2=A0 =C2=A0 =C2=A0Checking on this again, is this still a possib=

ility for merging into LMDB? This fix is still working great (improved pe= rformance) on our systems.

>=C2=A0 =C2=A0 =C2=A0Thanks,
>=C2=A0 =C2=A0 =C2=A0Kris
>
>=C2=A0 =C2=A0 =C2=A0On Mon, Jun 17, 2019 at 1:04 PM Kris Zyp <kris=

zyp@gmail.com mailto:kriszyp@gmail.com <mailto:kriszyp@gmail.com <mailt= o:kriszyp@gmail.com>>> wrote:

>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Is this still being considered/r=

eviewed? Let me know if there are any other changes you would like me to = make. This patch has continued to yield

>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0significant and reliable perform=

ance improvements for us, and seems like it would be nice for this to be = available for other Windows users.

>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0On Fri, May 3, 2019 at 3:52 PM K=

ris Zyp <kriszyp@gmail.com mailto:kriszyp@gmail.com <mailto:kriszyp@gma= il.com mailto:kriszyp@gmail.com>> wrote:

>
>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0For the sake of pu=

tting this in the email thread (other code discussion in GitHub), here is= the latest squashed commit of the proposed patch (with

>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0the on-demand, ret=

ained overlapped array to reduce re-malloc and opening event handles):

>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0https://github.com=

/kriszyp/node-lmdb/commit/726a9156662c703bf3d453aab75ee222072b990f

>

=20 =20 --=20 =C2=A0 -- Howard Chu =C2=A0 CTO, Symas Corp.=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0htt=

p://www.symas.com

=C2=A0 Director, Highland Sun=C2=A0 =C2=A0 =C2=A0http://highlandsun=

.com/hyc/

=C2=A0 Chief Architect, OpenLDAP=C2=A0 http://www.openldap.org/proj=

ect/

=20

--=20 -- Howard Chu CTO, Symas Corp. http://www.symas.com Director, Highland Sun http://highlandsun.com/hyc/ Chief Architect, OpenLDAP http://www.openldap.org/project/