[00:04:38] *** Parts: pwodkowx (~pwodkowx@192.55.54.36) ("Leaving") [00:05:53] *** Joins: pwodkowx (~pwodkowx@192.55.54.36) [01:56:01] *** Quits: pwodkowx_ (pwodkowx@nat/intel/x-pcoxbogarootawxz) (Quit: Leaving) [03:50:22] *** Quits: guerby (~guerby@april/board/guerby) (Read error: Connection reset by peer) [03:50:34] *** Joins: guerby (~guerby@april/board/guerby) [10:43:58] jimharris, peluse: does it make sense to allow zero-length xattrs in blobstore? [10:44:12] reviewing Cunyin's patch that adds an "is_deleted" xattr: https://review.gerrithub.io/#/c/374944/16/lib/blob/blobstore.c [10:44:26] I think it would make the most sense to not store any data at all, just use the existence of the xattr [10:44:51] for now, if the xattr API doesn't allow that, we could just make it store a byte but ignore its value [10:45:22] yeah - a zero-length xattr seems OK with me [10:47:45] also, I don't think blobstore.c should be modified for this - presumably the is_deleted xattr should be created and checked in blobfs [10:48:04] yes [10:48:07] absolutely [14:57:53] yup, no arguments here [16:11:15] drv - can you rebase bwalker's nvmf patch - you marked the first two patches +2 but looks like there are merge conflicts [16:11:25] "nvmf patch" => "nvmf patch set" [16:11:45] sure [16:12:34] I'm just going to rebase the first two for now - I need to review the later ones further [16:12:51] perfect [16:45:54] peluse: just noticed that I set the memzone_lookup mock thing to NULL, but maybe it should be set to MOCK_PASS_THRU_P? [16:45:57] (re https://github.com/spdk/spdk/commit/376c0de84d59d9fe19b8102311dd6e86cbac96d9) [16:46:14] I guess any other test that uses that stub will pre-configure it to the right thing anyway, so it probably doesn't matter [16:46:32] so FYI the developer meetup (see http://www.spdk.io/news/2017/08/21/dev_meetup/) is now full! please email me if you'd like to get on the wait list [16:47:21] drv, yeah, the PASS_THRU thing is really meant either for stubs that pay attention to it or for syscalls that are wrapped. For mocked funcs that are just "defined" with the macros just set it to the default that it was set to when defined (liek you did in that patch) [16:47:36] ok, cool [16:47:54] also, I should probably update the blog post (see above). Any preference on updating that exact post with a note or making a short new post? [16:48:03] we should probably just make a new post [16:48:07] either way works [16:48:13] 10-4 [16:48:38] for people who follow the RSS feed (probably just me), it is better to make a new post so it shows up again in the feed [16:51:02] we have an RSS feed? :) [16:51:21] maybe I should tweet it too LOL! [16:51:22] http://www.spdk.io/feed.xml [16:52:36] we should probably hook up a fancy icon on the website for it, but there is a link to it in the HTML, so pointing a feed reader at spdk.io should find it too