Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
Mateusz Warowny
@warownia1
Following the conversation we had some time ago: I'm studying the GNU GPL license and some forums regarding Python code. It seems that if you import GPLed library in your script/program then you must release it under GPL as well. They even explain it in FAQ. If the library is interpreted by an interpreter such as Java or Perl and you use this library in your code then the library is dynamically linked with your program thus you must release your program under GPL.
2 replies
Mateusz Warowny
@warownia1
I'm considering changing the style of the slivka configuration file. Here is the comparison of the old one and two suggestions. Which one you think looks better?
2 replies
b__soares
@b__soares:matrix.org
[m]
Elixir UK Hackathon Registration link: https://www.eventbrite.co.uk/e/second-elixir-uk-biohackathon-tickets-146597927347
benzoid
@benzoid
https://source.jalview.org/ seems to be down :( Gets as far as www-prod. Can't log in to see what's up as my VPN also not connecting.
Jim Procter
@foreveremain
Gitweb is working. The landing page to wwwprod is normal. What's broken ?
b__soares
@b__soares:matrix.org
[m]

I'm getting (last night and this morning):

% git pull
fatal: unable to access 'https://source.jalview.org/git/jalview.git/': The requested URL returned error: 500

The thing I said about www-prod can be attributed to late night rambling! VPN wouldn't connect (it would authenticate but not connect) last night but does now.

b__soares
@b__soares:matrix.org
[m]
Perhaps stale auth or something in the git clients.
b__soares
@b__soares:matrix.org
[m]
Okay my browser is doing better than my git client. It might just be me, although this is in multiple repo clones on multiple (2) machines.
Jim Procter
@foreveremain:matrix.org
[m]
not stale auth. my ultrasecret ssh pipe needs kicking .. did you do a restart of jv-atlassian ?
b__soares
@b__soares:matrix.org
[m]
nope, though got some info from Brian Russell will be useful (SMTP).
oh wait, maybe I did, but that was last week. Maybe I haven't used it since!
no... I did a git push yesterday evening which worked (I think). Better check that.
Jim Procter
@foreveremain:matrix.org
[m]
all works again
1 reply
Jim Procter
@foreveremain:matrix.org
[m]
@warownia1: b__soares
still in another seminar. will join at 3pm
Jim Procter
@foreveremain:matrix.org
[m]
morning all ! don't forget team meeting at 4pm today.
Mateusz Warowny
@warownia1
Hi all! Do you have some recommendations to how the server should re-try jobs that failed to start (i.e. node being down or some temporary issue). Currently I'm using a exponential backoff counter that increases the delay every time the job fails to start and resets when it starts successfully.
Advantage: not spamming qsub with requests that immediately fail and actually waiting for the issue to resolve
Disadvantage: Users need to wait several minutes until their job is finally marked as failed.
4 replies
Jim Procter
@foreveremain:matrix.org
[m]
Hi @benzoid - coming on for the developers meeting ?
b__soares
@b__soares:matrix.org
[m]
eek yes. sorry lost time again!
really need to sort out my meeting alerts!
Jim Procter
@foreveremain:matrix.org
[m]
see you all on Jitsi
foreveremain:matrix.org @foreveremain:matrix.org wonders what the plan is for today
Jim Procter
@foreveremain:matrix.org
[m]
@benzoid @warownia1 @arngucci I'm on jitsi if any of you drop by today
b__soares
@b__soares:matrix.org
[m]
argh, sorry -- having an "interesting" day -- delivery of 5m length of wood needed inspecting before the driver went off and then maneuvering into the right room. Got to dash now and pick up Brigid, then PDB50 starts! I'll fill in the doc later...
Jim Procter
@foreveremain
hi all - going to be 5 mins late! talk amongst yourselves :)
Mateusz Warowny
@warownia1
Hi all, what's the difference between state and status? Is it job state or job status? You say status code, but execution state. What's the difference?
b__soares
@b__soares:matrix.org
[m]
Hi @warownia1 , not sure I could put my finger on the precise difference! What kind of job do you mean? If it's like a computer process job I would talk about "process state" but for some reason I think I would say "job status". If I was talking about a working person's job, I would say "job status". I wonder if "state" is an instantaneous measure, e.g. "now", whereas status is a "now and ongoing" sort of thing? I might be making that up!
2 replies
Googling "job state" gets 505,000 results and "job status" gets 6,950,000 results, so the web says "job status" ;)
Mateusz Warowny
@warownia1
What metadata for the web service we want to present to the client? Do we need separate properties for things like: authors, URL, license etc. or can we shove all of that under the description property?
b__soares
@b__soares:matrix.org
[m]
I'm all for parseable metadata!
Is this relevant? https://www.researchobject.org/ro-crate/1.1/workflows.html
Some of the people we saw at the Elixir handson
Mateusz Warowny
@warownia1
They use @id and @type in the workflow description json files. Do you think it's a good idea to indicate some fields with @ or it's just confusing?
Also, when naming services and parameters I'm currently using a convention where name is an identifier (such as HTML input name) and label is a human-readable name. Should I change it to i.e. id and name?
b__soares
@b__soares:matrix.org
[m]
I'm guessing it's part of the json metadata markup. I'm afraid I haven't used json as a scheme much before -- it was all XML back in my days of metadata (well, once everyone realised SGML is impossible to parse reliably)!
I think it's okay for parameters of your service to be a separate thing from the labels of a metadata scheme. i.e. I don't think you should have to change. Is there a way of namespacing the parameters?
(or setting a default namespace for the parameters)
Mateusz Warowny
@warownia1

I think we are talking about two different things. I'm now re-organinsing REST API and I'm wondering whether I should change some parameter names. Currently, when you request e.g. the service you get the response like this

{
  "name": "<service identifier>",
  "label": "<human-readable name>"
}

I think it might be better to rename name to id, and label to name

b__soares
@b__soares:matrix.org
[m]
ah yes -- okay. If ypu can find/choose sensible looking pre-defined parameters from some other schema (the RO-Crates workflow might be a start, but there could be better ones) then it makes the service much more interoperable.
Mateusz Warowny
@warownia1
The names kind of made sense to me when I was thinking about it in terms of html forms where inputs have name that's not visible used to identify the field and label which is actually displayed. But then I realised that name and label might be a bit confusing. ro-crate uses @id and name
The @ they use may actually refer to JSON-LD
b__soares
@b__soares:matrix.org
[m]
If it's a machine-readable name then "id" makes sense.
b__soares
@b__soares:matrix.org
[m]
No Jalview team meeting today -- Geoff decision (Marek's birthday meeting going on)
Jim Procter
@foreveremain:matrix.org
[m]
Hi all - apologies for missing today's meeting - accidentally fell asleep !
2 replies
Jim Procter
@foreveremain:matrix.org
[m]
@arngucci @warownia1 @benzoid developer meeting today at 2.30pm uk ? Usual place meet.jit.si/JalviewDevelopers
b__soares
@b__soares:matrix.org
[m]
hee hee. I just naively googled "homebrew discussion" to ask a question about brew install and symbolic links. However I got lots of other interesting pages instead!! (although slightly disappointingly, the homebrew discussions page that I wanted was still the first result)
Mateusz Warowny
@warownia1
The first result in ddg is about brewing beer, wine, mead and cider, followed by AtariAge forum.
b__soares
@b__soares:matrix.org
[m]
:) something for everyone!
Robert Buels
@rbuels
:wave:
1 reply
b__soares
@b__soares:matrix.org
[m]

Jim Procter: I've been processing the Jira users and Discourse users (i.e. the mailman users). I'm happy to go ahead with
a) Making changes to username, name and jira-developers=>jalview-dev-me mapping in Discourse to match those already existing in Jira. This affects 58 Discourse accounts;
b) Creating accounts on Discourse for Jira users whose email address has not been used in mailman. This would be set up to happen automatically anyway when they log in, but they wouldn't get any alerts until then (so wouldn't necessarily know to do so). This creates 352 Discourse accounts [seems quite a lot-ish];

I want to check with you about:
c) Creating 435 Jira accounts to match accounts on Discourse that do not match an existing Jira account.

I think the 352 in step (b) seems like quite a lot, but maybe not. There is information about the user's last login, and I could limit to accounts that have been used in the last 10 years, but even then I can imagine a user who reported a bug 11 years ago on the issue tracker, and then didn't use the issue tracker again but does still use Jalview, so maybe no limiti is necessary.
I don't have any last login information for the Discourse (mailman) users so couldn't limit step (c) like this anyway.

Also, I have got my adapted discourse-crowd plugin working and configured so that when a user logs in with a Jira account in the jira-developers group then they automatically get added to the jalview-dev-me group. If they log in without being in the jira-developers group then they get removed from the jalview-dev-me group. It's a fork in our jalview github, I should probably send a PR at some point. Note that this only happens when they log in, so I am pre-populating the group in step (a) or they won't be able to carry on using jalview-dev@jalview.org as a mailing list (until they log in on the web). Also note that membership is only revoked when they log in, so we could have a cron job or something to do this.

Let me know what you think about (c) [it might be "just go ahead", which is what I'm veering towards].

2 replies