MessageSvcSink: Increase default histogram string width in printout

Current width is too short for a number of histograms in various LHCb tests. This can lead to ambiguity on the histogram ID, e.g.

https://mattermost.web.cern.ch/lhcb/pl/9yzfaa5oqtdz8rybnrzdwc4ekw

 | TAE/MainEvent/perModule/LiteClustersPerChanne | "Lite Clusters Per Channel in T2VQ1M0;Cha..." |        184 |     1019.6 | 567.86     |  -0.036134 |    -1.1125 |
 | TAE/MainEvent/perModule/LiteClustersPerChanne | "Lite Clusters Per Channel in T2VQ1M1;Cha..." |         64 |     1139.5 | 549.62     |   -0.47605 |   -0.73359 |
 | TAE/MainEvent/perModule/LiteClustersPerChanne | "Lite Clusters Per Channel in T2VQ1M2;Cha..." |         50 |     771.04 | 505.73     |    0.49597 |   -0.44699 |

See how the truncated IDs for all these are the same. I also know of a lot of example in the RICH and other sub-systems where the same happens.

This leads to issues in the testing.

@sponce FYI

If we're anyway touching references downstream, why not (also) move the ellipsis in the middle of strings that don't fit the configured width?

I don't get what you mean. That just indicates in the title string it is truncated. I don't really know why it does this only for the title string, and not the IDs, but thats for @sponce to comment on.

I mean to "truncate" in a more useful way so that collisions are less likely:

Instead of this

• ThisIsAHistogram12222 -> ThisIsAHistogram1...
• ThisIsAHistogram13333 -> ThisIsAHistogram1...

do this

• ThisIsAHistogram12222 -> ThisIsAHi...ram12222
• ThisIsAHistogram13333 -> ThisIsAHi...ram13333

Ah, I see what you are suggesting now.

The thing is , for histograms, the field isn't just the title, it is actually

title;x-axis-title;y-axis-title

so its not clear to me if doing what you suggest is best for titles.

For the IDs, yes, it makes sense I would say. There nothing is currently done so it should be added.

I think though this is all beyond the scope of this trivial little MR and something for @sponce to ponder.

I was curious so see !1704 (merged)

I have nothing against increasing the default here, but you will never fit all possible cases, only moving the tail further. So the actual solution for the few cases exceeding the default size is to set the Property HistoStringsWidth explicitly in the test options. Do we have so many cases that we need to change the Gaudi default or can we do this in the few problematic tests ?

This MR is largely moot now, for LHCb, as I have made the effective default as set by PyConf bigger. So do as you wish here.

This is indeed a good alternative. I would then close this Gaudi MR. What do you think @clemenci ?

mentioned in merge request lhcb/LHCb!4959 (merged)

mentioned in merge request lhcb/Rec!4283

mentioned in merge request !1704 (merged)

closed