2015 06_Meeting_IRC_Log

Michal Čihař edited this page Apr 14, 2016 · 2 revisions
Clone this wiki locally
<Marc9> Item 1: supporting 30K tables in one db
<madhuracj> Hi everybody
<Marc9> I would say to that: it's an edge case, we won't support this
<dstorm> Does MySQL has any restrictions on it?
<madhuracj> 30k looked little too big
<Marc9> dstorm: no
<dstorm> then I would say why are we blocking users to do so?
<Marc9> dstorm, we are not blocking
<zixtor> What seems to be the issue btw? If someone researched..
<dstorm> I mean if it stops responding then its like they are blocked to do so
<Marc9> The request never completes
<zixtor> dstorm: its not that we are imposing it consciously ;)
<zixtor> anyway it surely seems an edge case
<madhuracj> According to the reported everything works
<madhuracj> just that it take time
<dstorm> Can using legacy navigation in that case does any good?
<Marc9> dstorm probably
<madhuracj> Legacy navigation is just UI on the same loading mechanism
<madhuracj> So I suppose it wont help
<zixtor> I think it will make only minor difference
<Marc9> dstorm, you mean pre-4.0 ?
<dstorm> Marc9, yes
<madhuracj> I was talking about the configurable feature we have in 4.4
<dstorm> I always had concern with the performance issues of new navigation tree.
<Marc9> Anyway, if we agree that it's an edge case ...
<dstorm> I agree that its an edge case
<madhuracj> Of course I agree
<dstorm> but if there is a way to fix it and we should look for it
<zixtor> I too
<dstorm> *then
<zixtor> I think it would require much effort to further optimize existing navigation mechanism
<Marc9> dstorm I believe we have other priorities which are not edge cases, so I would close this "won't fix"
<madhuracj> I can investigate the issue and write to the devel mailing list whether the situation can be improved
<Marc9> madhuracj sure but don't lose too much time on this, please
<madhuracj> Sure.
<Marc9> Moving on?
<dstorm> Marc9, I am not saying that we should fix this in priority but we can come back to this later
<dstorm> We can move on then
<Marc9> dstorm, well, we said that for other bugs and they are in the tracker 12 years later
<Marc9> 2. Bring back the start row and number of rows in Browse
<madhuracj> I do not see a strong reason to bring it back
<Marc9> Isaac said that he is not in favor; me neither
<zixtor> I also feel having such feature does amount to cluttering the interface..
<nijel> I also don't see much need for this
<Marc9> I am happy with the uncluttering that removing this feature did in the past
<dstorm> I also feel that it won't be used much
<Marc9> Moving on
<Marc9> 3. Browse mode: apply functions to columns (SUM, AVG, etc)
<Marc9> 11 years ago, Michal was unsure about this being used enough to be justified
<Marc9> I share his view
<Marc9> Opinions?
<zixtor> I find it may be useful but again many things can be useful..
<nijel> I haven't changed my opinion in 11 years ;-)
<dstorm> Maybe this could be added in query builder
<madhuracj> I share the same view
<Marc9> zixtor right and we must think about keeping / reaching a clean UI
<Marc9> dstorm IMO it's already there in the query builder, but
<Marc9> the visual query builder is not well known
<Marc9> Moving on
<dstorm> Marc9, to be frank even I have not used visual query builder ever
<Marc9> 4. Working around a PHP bug
<nijel> no workaround for php bug
<Marc9> Especially since this PHP bug has been fixed
<madhuracj> exactly
<Marc9> It's not really our problem if distros are slow to integrate fixes
<zixtor> Agreed..
<Marc9> Also, we did not hear about this problem much
<Marc9> Moving on to our "big" discussion item
<Marc9> 5. Unit tests and HTML
<Marc9> zixtor we are listening to you :)
<Marc9> (then to others)
<zixtor> Well, as we shift to template system, changes in HTML will mostly be just format changes and they will mostly deviate from the static assertions in unit tests..
<zixtor> So in my view its a waste of time to keep format correcting the HTML in these tests... and they come of use rarely
<Marc9> Is it ok to say that we have functions that generate HTML and functions that do not generate HTML? ...
<zixtor> Same utility can be derived by checking for keywords and just having html validation..
<Marc9> So, we should rethink our tests for functions who generate HTML
<dstorm> Even, I don't think we should assert HTML output in unit tests and instead assert only the data that data is being put into it.
<zixtor> Marc9: surely we have both kind of functions
<dstorm> *assert only the data that is being put into it.
<Marc9> Note that the removal of assertTag in phpunit gives us even less choices
<zixtor> Marc9: right
<zixtor> Maybe we can change to data checking as we templatize a particular script.. and till then keep it same..
<Marc9> I mean, it's one thing to test a function like pow(), and a function that generates HTML
<Marc9> zixtor right
<zixtor> yeah they are quite different
<Marc9> For example, a function that is supposed to generate a dropdown (SELECT, OPTION): chances are thin that it would fail
<dstorm> I think Selenium is the correct place for UI tests.
<zixtor> Marc9: agreed..
<zixtor> dstorm: right
<Marc9> zixtor, is there an example of data checking in tests done by Jason?
* udan11 a quitté (Remote host closed the connection)
<zixtor> Marc9: no not yet
<zixtor> maybe we can ask him to start working on that
<Marc9> and did he give his views about unit testing his code?
<zixtor> So far that there should be something other than asserting for html.. no idea as he seems to be working further leaving tests for now
<Marc9> dstorm, with Selenium won't we have the same problems as asserting HTML output?
<Marc9> (on moving code)
<dstorm> we won't be asserting html outputs but instead doing user actions to ensure elements are behaving a they should
<dstorm> so if HTML breaks then things like dropdown, checkboxes should stop working
<dstorm> *as
<Marc9> dstorm, OK, I'm not familiar enough with Selenium
<Marc9> zixtor, what would be the new way of testing <input type="hidden" name="field" value="foo" />
<zixtor> dstorm: do we have selenium coverage of most UI? I am not familiar abt it
<dstorm> I don't think it covers most of the UI.
<Marc9> Wasn't there a problem with Selenium on Travis?
<zixtor> Marc9: each template right now uses some data variables.. so we can assert that whole HTML contains valid values of those variables and we can just validate the whole HTML..
<Marc9> zixtor ok
<dstorm> assert the data and just parse the html to see if there is DOM error?
<Marc9> nijel madhuracj anything to add on this?
<nijel> dstorm: that sounds good enough
<Marc9> By the way, via email Isaac said that he also doubted the usefulness of testing HTML
<madhuracj> Well, I agree it's little point asserting html tags
<zixtor> I just saw a popular service to validate, https://validator.nu/ we can research more
<Marc9> zixtor can it be linked to phpunit? I guess not
<dstorm> we can use XML parser in php?
<zixtor> dstorm: but yeah we can just parse it..
<Marc9> To conclude, we won't add more HTML tests and as we move to templates, we won't assert HTML tags
<Marc9> Can we briefly discuss the next IRL meeting?
<madhuracj> Sure
<zixtor> ok
<Marc9> We had suggestions for March 2016 and summer 2016,
<Marc9> and I added suggestions for October 2015 and August 2015
* cipi est maintenant connu sous le nom CiPi
<Marc9> So, it 2015 too early?
<Marc9> is
<zixtor> I expect I will be able to attend those in 2016.. But for this year, in August or October it wil be difficult to find time for me..
<madhuracj> August 2015 would not be possible for me, but others I would be able to attend.
<Marc9> Ideally we should aim to have a maximum of participants
<dstorm> August 2015 seems too early
<Marc9> I guess that Singapore was the most promising one (March 2016)
<dstorm> Maybe we can do vote similar to what we did for choosing the time for team meeting which would give us an idea when maximum team members would be available
<zixtor> Surely sounds like most members can attend that..
<madhuracj> Marc9 Can we do something like a poll where everyone can rank conerences based on availbility and on how interesting the conference is
<madhuracj> I think we did that for previous IRL meeting
<nijel> madhuracj: sounds like a good idea to sort them
<Marc9> madhuracj sure, I'll organize this, but is it too soon to ask for 2016? I mean, we don't even have the dates for DebConf 16
<zixtor> yeah will be nice to have vote/survey to have most people attend as its quite worth it to meet everyone after traveling ;)
<madhuracj> Marc9 true
<Marc9> what's the delay for you Asian guys, about the visas? 2 months?
<dstorm> Marc9, lets add only those options for which dates are finalized and later add more options?
<Marc9> dstorm, I would prefer to do a final poll with a fixed limit date to answer
<dstorm> Marc9, depends on where we would be going
<zixtor> It can be applied for Europe 10 days before, i think Smita got it this year
<madhuracj> But 2 months should generally be enough in my experience
<dstorm> I applied very early this time but was bit troublesome for me.
<Marc9> Ok so a poll at beginning of 2016 should be fine
<Marc9> dstorm you were early and got troubles?
<madhuracj> Whether each member needs to obtain a visa to attend each conference can be a question in the poll
<Marc9> madhuracj good point
<Marc9> Anything to add on this or other subjects?
<zixtor> Still 8 minutes to go
<dstorm> Trouble as in had to go for an interview to consulate which was quite a journey
<Marc9> dstorm maybe Singapore will be easier in this case
<dstorm> Yes, it would be for sure.
<madhuracj> I do not require a visa for Singapore
<madhuracj> I suppose it is same for dstorm and zixtor
<Marc9> and Smita
<madhuracj> yes
<zixtor> yeah need to look abt that on net
<Marc9> I do not require a visa for Singapore
<zixtor> Yeah it seems its available on arrival
<nijel> I don't seem to need it as well
<Marc9> With this, I hereby declare:
<Marc9> *** End of team meeting ***