fishtest
fishtest copied to clipboard
glinscott
add a few more approvers
right now, looks like we're short on approves, and I think some regulars really would be happy to help out. Just looking at https://github.com/official-stockfish/Stockfish/graphs/contributors?from=2017-01-01&to=2018-09-29&type=c people like @protonspring @Rocky640 @joergoster @candirufish @miguel-l would be obvious candidates.
I sm happy to help out if you give me some basic rules.
On Sun, Sep 30, 2018, 2:11 PM Joost VandeVondele [email protected] wrote:
right now, looks like we're short on approves, and I think some regulars really would be happy to help out. Just looking at https://github.com/official-stockfish/Stockfish/graphs/contributors?from=2017-01-01&to=2018-09-29&type=c people like @protonspring https://github.com/protonspring @Rocky640 https://github.com/Rocky640 @joergoster https://github.com/joergoster @candirufish https://github.com/candirufish @miguel-l https://github.com/miguel-l would be obvious candidates.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/glinscott/fishtest/issues/330, or mute the thread https://github.com/notifications/unsubscribe-auth/AcVnSBsY1v2jYz_L8h4buTmmy0_mEhQxks5ugSWPgaJpZM4XBIkW .
The first check is to make sure the patch contains no malicious code. Basically, approving is the only guard against that. After that, enforcing common sense and spotting easy mistakes ... are the bounds correct (e.g. [-3, 1] misuse, [0,4] oversight..), and tuning setup reasonable. Right now resources available are abundant, but one day this might change, if so suggest people that overuse the system to adjust e.g. throughput, set priority, or just ask to delete the patch.
Sounds fair. I am happy to help out.
On Sun, Sep 30, 2018, 2:18 PM Joost VandeVondele [email protected] wrote:
The first check is to make sure the patch contains no malicious code. Basically, approving is the only guard against that. After that, enforcing common sense and spotting easy mistakes ... are the bounds correct (e.g. [-3, 1] misuse, [0,4] oversight..), and tuning setup reasonable. Right now resources available are abundant, but one day this might change, if so suggest people that overuse the system to adjust e.g. throughput, set priority, or just ask to delete the patch.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/glinscott/fishtest/issues/330#issuecomment-425748732, or mute the thread https://github.com/notifications/unsubscribe-auth/AcVnSGeGRmwtTvapTQdj_fxoCQ0ITi5Vks5ugScwgaJpZM4XBIkW .
There are no active tasks, but some in the queue. I tried to approve one, but it says I need to login (which I already did). Am I supposed to be able to approve tests now?
On Sun, Sep 30, 2018 at 2:26 PM Michael Whiteley [email protected] wrote:
Sounds fair. I am happy to help out.
On Sun, Sep 30, 2018, 2:18 PM Joost VandeVondele [email protected] wrote:
The first check is to make sure the patch contains no malicious code. Basically, approving is the only guard against that. After that, enforcing common sense and spotting easy mistakes ... are the bounds correct (e.g. [-3, 1] misuse, [0,4] oversight..), and tuning setup reasonable. Right now resources available are abundant, but one day this might change, if so suggest people that overuse the system to adjust e.g. throughput, set priority, or just ask to delete the patch.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/glinscott/fishtest/issues/330#issuecomment-425748732, or mute the thread https://github.com/notifications/unsubscribe-auth/AcVnSGeGRmwtTvapTQdj_fxoCQ0ITi5Vks5ugScwgaJpZM4XBIkW .
-- Michael T. Whiteley mobile: 801.707.6886
I'm commuting now. I will be able to grant the approvers right in a couple of hours.
It's not "protonspring" ?
On Mon, Oct 1, 2018, 12:43 PM ppigazzini [email protected] wrote:
@protonspring https://github.com/protonspring I need your username in fishtest.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/glinscott/fishtest/issues/330#issuecomment-426018114, or mute the thread https://github.com/notifications/unsubscribe-auth/AcVnSK-dWJF56ZAfgFrFayl7ruOfuXriks5ugmJKgaJpZM4XBIkW .
@protonspring yes, I misspelled the username when searching in the db, sorry :)
@Rocky640 @joergoster @candirufish can you also leave your fishtest username here if you could help?
I also would like to apply for an approver role. Basically now I'm the most active fishtest user and see quite a lot of the time that good tests are pending (especially @vondele ones). Sure, I'm not that experiences in coding yet but I wouldn't approve obvious garbage :)
@Vizvezdenec I will glad to grant you the approver right as soon as your candidature will have the OK from an experienced SF developer. I'm not active on SF development.
@Vizvezdenec ... I do appreciate a lot your active patch development, but in the list above I tried to restrict it to people that have already quite a few patches committed to trunk. I know that limit is a bit arbitrary, and I think we should encourage new people to join the project, but for now I think it is better to restrict approver rights somewhat.... I do hope that you get there sooner than later, however.
Personally I feel @Vizvezdenec is a perfect candidate. He may not have many patches in (and neither did I when I was given approver), but I know for a fact he reads through everything, checks against old patches, and does the simple sanity things about TC/settings/bounds that we need. With so many approvers accepting literal garbage, something like @Vizvezdenec is a valuable addition.
I didn't realize that we needed to check against old patches, but I'll start doing a cursory search. I've mostly been approving it the code looked like a legitimate attempt at something and wasn't broken in some way. I'll be a bit more careful in the future.
On Thu, Oct 4, 2018 at 1:59 PM Andrew Grant [email protected] wrote:
Personally I feel @Vizvezdenec https://github.com/Vizvezdenec is a perfect candidate. He may not have many patches in (and neither did I when I was given approver), but I know for a fact he reads through everything, checks against old patches, and does the simple sanity things about TC/settings/bounds that we need. With so many approvers accepting literal garbage, something like @Vizvezdenec https://github.com/Vizvezdenec is a valuable addition.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/glinscott/fishtest/issues/330#issuecomment-427149672, or mute the thread https://github.com/notifications/unsubscribe-auth/AcVnSPgzSGUmH_W1cKONTox991GACtraks5uhmiGgaJpZM4XBIkW .
-- Michael T. Whiteley mobile: 801.707.6886
If this issue is still under consideration, I would like to recommend that @Vizvezdenec be given an approver role.
At the moment, approval delays for tests are predominantly determined by the time of day the test is submitted; some times have many approvers and others have very few. My own experience, as someone who is predominantly active during the night (eastern U.S. time), is that it is still common for tests to wait hours for approval. I think many people who are not active during these times (including, obviously, many approvers) understandably do not realize how long these delays are.
I recognize the efforts of @mstembera and @protonspring; each of them has helped this situation significantly. During these hours, the vast majority of the time (I would guess at least 85%) it is one of them that approves patches from users such as @Vizvezdenec and myself. Before they became approvers, delays were much longer, and I'm quite grateful to them for the valuable hard work they have donated to this project.
But I don't think only two people should have the burden of often handling all approvals themselves for almost a quarter of the day, sometimes longer. If you look at http://tests.stockfishchess.org/actions at the moment, for example, you can see that for 7.5 hours and counting, no other approver has approved a test. This is not at all unusual, and I think this is too much to ask of only two people. We need another approver who is active during these hours.
I believe @Vizvezdenec is that approver. Not only is he active when we need approvers the most, but he is also highly qualified. He understands SF code well and (as @AndyGrant noted) diligently performs all the checks we need. I believe he is highly capable and his work as approver would dramatically improve wait times.
I'm not sure who specifically is responsible for determining who is an approver, but I urge you to consider granting @Vizvezdenec this well-deserved responsibility.
Given that indeed we're still having to few active approvers, and the recent activity of @Vizvezdenec I'm also fine with giving him the additional responsibility.
Saying that I know code really well is a bit of stretching, but I will do my best :)
honestly I propose @31m059 as an approver. Because sure, I'm approving tests in dead times but my tests remain stalled for hours and it's kinda sad to see. @31m059 will close the worst time in terms of making tests + he is quite experienced in SF code and writes quite a few patches frequently so is in touch with all latest changes. Of course if he wants to be one :)
@Vizvezdenec I noticed your patches stalling recently and was actually considering posting here to volunteer when I saw your comment.
Thank you for the kind words. I really appreciate the time you've put into the "night shift" at fishtest but agree that we still don't always have enough approvers. I would be more than happy to help!
Recently my tests are idling for quite some time even fishtest is not under any load. So I propose to add @MJZ1977 and @ElbertoOne to approver teams as one of the most active devs lately w/o right to approve and both are experienced and knowlegeable (if they are not against this, of course) @ppigazzini
Thank you @Vizvezdenec for proposing my name. If it can help, I accept to be approver. I hope that I will not make mistakes and just precise that I will let other approvers make the job for tuning tests because I did'n have experience on these tests.
If all agree that I can be an approver, then I'm willing to accept this role.
http://tests.stockfishchess.org/tests/user/MJZ1977 http://tests.stockfishchess.org/tests/user/ElbertoOne