We just deployed a new point upgrade to rehash today to fix a bunch of small bugs that have been with us since the rehash upgrade and a few that were around longer than that. Here are the highlights:
We were able to kill off about 10 high priority bugs with this mini release. Current issues and feature requests can be found on GitHub and you can submit new issues or feature requests here if you have a GitHub account. We also welcome bugs and requests sent via email to admin@soylentnews.org or left in the comments below.
Our goals for the next major update is more of the same bug hunting and killing with a few features added here and there. Again I would like to thank you for your patience with the growing pains we have had with the 15_05 rehash upgrade. This update should bring us mostly back to where we were before in terms of broken features.
(Score: 2) by tynin on Tuesday June 16 2015, @09:47PM
I LOL'd.
Nice round of fixing and show of effort to get things back to sailing smoothly so quickly. Still, I wish you all better luck and making future releases more seamless if only for your own sanity. :)
(Score: 4, Informative) by paulej72 on Tuesday June 16 2015, @10:04PM
We basically decide that we were never going to get rehash out the door if we waited to test and debug. Most of the major issues did not show up under the light load of the dev server and would have passed any additional testing we may have done. Also there seems like we have some schema differences between our dev db and production db. This also led to a few issues.
NCommander plans to work on getting some unit tests built for each of our Perl modules and files. Once completed this should help with keeping the bug count down. As part of this we will also be killing off some more dead and dying code in rehash. This will make it easier to keep the code up-to-date and bug-free.
Team Leader for SN Development
(Score: 2) by goodie on Wednesday June 17 2015, @01:55PM
RE: differences between dev and prod dbs: How did this happen? Didn't you test by taking a backup of prod and putting it in dev running only those scripts that were for the new version? Anyway, a good diff tool for database might come in handy for the future to check whether differences exist before doing the upgrade. My old company had SQL Compare for MSSQL, which could even generate a convert script one way or the other. For MySQL, looks like there is some stuff built-in too: http://dev.mysql.com/doc/mysql-utilities/1.3/en/mysqldbcompare.html [mysql.com] (not sure about the version). Or a simple script that compares two databases' catalogs could do that for you if you have basic objects only.
(Score: 3, Interesting) by paulej72 on Wednesday June 17 2015, @05:15PM
The current dev database is a copy of the production db from around last year at this time. Since then we have added vars and made schema changes as needed to provide for new features and bug fixes. These changes are supposed to be written to the sql/mysql/upgrades file to then be applied to production at the next deploy of the code. The problem arises when someone makes a change to the db but fails to add the change into the upgrades file. So we have both schema differences and some vars in the configuration table that are slightly different.
What we will probably do soonish is take another dump of the prod db and put it on dev to give us the same db again. Or we might just dump the schemas on both and do a diff.
Team Leader for SN Development
(Score: 2) by goodie on Thursday June 18 2015, @01:22PM
Thanks for the info! Maybe running a diff before upgrades could help if you have a code freeze period where you can validate those things. Because I imagine that it's the same type of issue that can arise with server configs etc. since none of that stuff is "built" and therefore fails to compile etc.
(Score: 2) by mcgrew on Wednesday June 17 2015, @07:39PM
Maybe I just haven't been paying attention, but I haven't noticed any bugs or annoyances. I do have a suggestion, though. This would do me no help since I never log in on my phone, but I was curious and looked at the site on it. You really should make it mobile-friendly. It isn't hard using CSS (I don't on my site... much) and you can test it here. [google.com]
Google now reduces pagerank for sites not mobile-ready. Despite the fact that I get almost no traffic from Google, my stats say that before I changed the code, very few people were reading on a phone or tablet. Now half my traffic is.
We not only don't have all the answers, we don't even have all of the questions.
(Score: 2) by paulej72 on Wednesday June 17 2015, @07:52PM
That has been on our todo list for a long time, even though it is not listed as an issue on GitHub we are well aware of it. Our Chief QA man martyb uses a phone for internet access and has mentioned many mobile browser related bugs.
Team Leader for SN Development
(Score: 0) by Anonymous Coward on Thursday June 18 2015, @07:44AM
But please, refrain from making separate URLs for a mobile version. Such separate URLs are a PITA because you can bet that the URL you get is typically not the one appropriate for the system you're on. It sucks to view a mobile-only-optimized page on a big 1920x1080 screen.
(Score: 2) by paulej72 on Thursday June 18 2015, @12:52PM
Well it also sucks to get stuck on the mobile version without having a way to set it desktop view the way some responsive layout systems work. I like responsive layouts, but it would be nice to be able to set a cookie that allows the user to turn it off. My phone screen may be small, but it is high resolution so can handle a full desktop version if needed.
Team Leader for SN Development
(Score: 2) by juggs on Thursday June 18 2015, @05:02AM
Indeed so. Any volunteers with QA/Test experience would be most welcome, don't sit on the sidelines, get involved.
To be fair, rehash deployment was a major^2 step. Given our skeleton crew there were bound to be gotchas, we simply do not have the resources to iron everything out in dev / staging right now.
The rapid march of point releases we have seen since rehash deployment attests to the commitment that the developers have.
slashcode to rehash was always going to be a breaking change as it touched on every underlying platform. It's done now and we have a current codebase and platform to work from. Sure there are regressions to deal with, but this was a sand or stone foundation choice, one better made sooner than later. Despite the recent brief site instability, it was the right way to go.
/my opinion off