Square3dreamreal, oh. It's that's much of an upgrade?
dreamrealhikari's pretty good, yes
dreamrealif c3p0 is working fine for you, eh, stay with it, I guess, but I'd do the migration personally
deebohikari has been pretty much "set min connections to production mean, max bit above that, done"
deebothere's also a "secret" system prop to have it initialize up to min connections before returning from the ctor, if you're deploying to any existing load
kcomhnalldreamreal: hey are you around?
kcomhnallwondering if you resolved that forgotten windows password issue yet.
dreamrealI guessed it was hunter2 and I was right
dreamrealI have a crapload of things to write and no passion to draw on for them. :/
kcomhnalldreamreal: heh... I had also forgot my own. turns out setting a password expire date forces a new password entry.
kcomhnalldreamreal: any of that crapload can be offloaded or relegated?
dreamrealWell, the idea was that THE COMMUNITY would create drafts, based on what THE COMMUNITY thought was interesting
dreamrealthat's why submissions can be made via IRC, or anonymously (no login required), etc
kcomhnallfor specific projects of just in general sense?
kcomhnalls/of/or
dreamrealgeneral sense
dreamrealI actually have: graphcompose, openliberty and jee11 (they have a really good 'catch up on 11' post), endive.run, apache fory
javabotdreamreal's title: "GraphCompose 1.6.4: A Fluent PDF Generator | bytecode.news"
ParaWhy is reading PDFs programmatically so bad.
ParaI'd probably be better off turning this PDF into a series of images and doing a crap OCR on the text data than trying to directly extract it :P
dreamrealPara: because PDFs aren't meant to be *read*
dreamrealtika can help, although tika sucks in a lot of ways
dreamrealpandoc may also be able to help
cheeserPara: icepdf?
Parahm
cheesersorry. pdfbox.
cheesereither would probably work but pdfbox is oss.
ParaI've had a poorly working pdfbox already. Would probably need to try again.
dreamrealWhat are you reading the pdfs FOR
cheeserooh. i just found openpdf from the libreoffice people.
cheeserthat's likely very good given the source
Paradreamreal: The Finnish radio spectrum allocation table which is distributed as pseudotable PDF for some reason.
dreamrealurgh. For tables, rendering to OCR might be a good option, actually
cheeseri used itext way back in the day but they switched to GPL and that killed it for me.
ParaAnd the governmental body is basically going "whatever this matches the legal requirement for open data"
ParaAnd i'm 100% sure the actual source for this PDF even is CSV/some other columnar flatfile format which would be tons better.
BombeEh, it’s probably a Microsoft Word document…
dreamrealI really wish docbook was the lingua franca
dreamrealbut nobody wants to type XML
cheeserLaTeX for life!
dreamreallatex doesn't have all of the features it needs, though
dreamreallatex is for typesetting, documents are not just typesettings and content
Swayzeor to prevent life!
NeXeNxml isn't bad really. i never type it i always parse it
NeXeNif you pack 2000 records into xml and pass it, it is probably pretty verbose unless you name things like <s> and <a> and <d/>
jreicherDoes it make sense to do XML without a DTD though? That's what I've always wondered, and people generally don't bother with DTDs
jreicherWell, not DTD specifically I guess. A schema of any kind.
ernimrilNeXeN, uh? xml is slow to parse and uses a lot of memory and it not readable by humans. In what way is it not bad?
NeXeNif you use it for very small things it's ok. like html for instance
NeXeNit's usually the fastest thing on the page, because the browser optimizes it. it's equally good at semantic embeddings for nlp, which are easy to parse (though they may be slower than say raw data in a protobuf
NeXeNbut it's definitely not a good data transport or persistance architecture
NeXeNmany people don't like it for config either but i use my pom every day. annoying, yes. impossible, no