Kulmegil

Member
  • Content count

    175
  • Joined

  • Last visited

Posts posted by Kulmegil


  1. Valkunt starting 0.3.10 works fine with new dc++ based clients (and ApexDC++) both ways.

    ShakesPeer 0.9.5 for sure (don't remember if older had own dctag)

    SDC++ 1.00 RC9..

    PS Generally I banned all clients older then DC++ 0.69 & 0.68 (with small exception for 0.674) based. And I still think I was to tolerant here :) DC is P2P network and not museum. Besidedes it's good for overall both security and compatibility imho.


  2. Just don't ask. There are other peoples in the queue waiting to get the slot also (and for sure I wouldn't like if each user ask me for slot every time he has to wait). Resources like upload are always limited. So stay put and wait for Your turn.


  3. Jumping from hub to hub is not the way for DC network. You pick Your best hubs with users having lots of good stuff and stick them (and search some new when necessary). If You want to find some rare file just go to some content oriented hub or use networks like eDonkey with much more "global" serach options.


  4. /me vote no, coz me don't want situationes when useres are able to suppress chats and block all pms (next step) and becoming like unresponsive blockheads just taking space on userlist. :)

    Not that I don't see the point in You're arguments... but DC is like P2P & irc-like network in 1. It wouldn't be half as popular if it would be just one of those. So dooon't change it!


  5. But I'm thinking that some ...let's call it that... negligence was made in the past. Why don't we just define something new like $utf8 - something optionally sent by the hub at the very beginning when user is connecting (much like $HubTopic). It say that utf-8 encoding should be used on this hub. It's incompatible? SO WHAT!? Obviously all saying that ADC is next door and it will support utf-8 and it's gonna be cool are right... but if it would be implemented when first Unicode aware DC++ showed up, I'm pretty sure we would be living in Unicode version NMDC today. A shame imho. :(

    There are obviously more fields where breaking some compatybility could lead to better DC world already today.

    But the good thing is since that damn ADC is really close much of it will be solved at last.


  6. Hunspell (OOo, Miranda, .. Firefox3?), or Aspell are probably the best choices currently. I don't know witch is better and I wouldn't dare to open discussion about it myself (both are open source, UTF8, have lot's of dictionaries...), but probably the one witch is easier to implement. I suggested miranda plugin because I count it would be easiest (convert it to something like peerguardian plugin?). The more difficult it gets the less chance to see it (this century at least:)

    As for the resource usage I already posted my observations for IM plugin (uses Hunspell). Impact on memory looks minimal if You don't use the feature.

    Whatever the way, it's goal that meters : it's suppose to do spellchecking "as I type", underline bad words and give context menu suggestions. It's the only reasonable way for IM style conversations.


  7. Main purpose of DC is p2p (sharing), chat is only minor feature. SpellChecker would only decrease DC usability (higher memory usage, higher CPU usage etc.).

    If there should be some spellchecker, it should be implemented as a plugin in hubsoft. For example, some czech hubs warn you via ChatBot if you make some grammatical mistake ;-)

    I would say 90% is for P2P and 10% for chat when I was starting. But today it's 50/50 if not less.

    As for memory usage...

    virtual mem / peak mem usage

    16 384 kB / 21 596 kB with spellcheck (2 dictionaries)

    10 508 kB / 16 764 kB with spellcheck (no dictionaries)

    10 292 kB / 16 672 KB no spellcheck loaded

    0.7.0 build #30 Unicode, 24 total plugins, after start - messageing window opened.

    I would say it's worth considering at least (depending on difficulty of implementation).

    It could be implemented as plugin (like guardian plugin in Zion++)

    PS Spellchecking in hub is evil (except badword checking and spam checking) - should be done client side imho.