napshare-developer Mailing List for NapShare (Page 3)
Brought to you by:
web1
Menu
▾
▴
napshare-developer — Developer Chat
You can subscribe to this list here.
| 2001 |
Jan
|
Feb
(2) |
Mar
|
Apr
|
May
(1) |
Jun
(2) |
Jul
(3) |
Aug
(1) |
Sep
|
Oct
|
Nov
|
Dec
(1) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2002 |
Jan
(7) |
Feb
(16) |
Mar
(21) |
Apr
|
May
(2) |
Jun
(1) |
Jul
(1) |
Aug
(2) |
Sep
|
Oct
|
Nov
|
Dec
(1) |
| 2003 |
Jan
(1) |
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(1) |
Nov
(3) |
Dec
|
| 2005 |
Jan
|
Feb
|
Mar
|
Apr
(12) |
May
|
Jun
(2) |
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2006 |
Jan
(2) |
Feb
|
Mar
|
Apr
|
May
(1) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: Nate <we...@mo...> - 2002-02-27 09:14:13
|
Remember, once you get anon CVS going you only have to do a cvs update and to back out to a previous version, or just want a fresh copy, remove all files except CVS folders and do cvs update -D "two days ago" or cvs update -D "02/20/2002" This update adds the "and" filter, uploads show user agent, debug in auto shows filter now, and who knows... Oliver, I tested the or filter and it seems to work correctly along with the size, so please turn on auto debug, log to a file with > log.txt, it will print if it queues or not + what filter kicked it out and you will be able to tell why and send me some sample strings to try out here if it fails. Thanks for the test reports! Keep'em coming! Enjoy! |
|
From: Oliver T. <ol...@bo...> - 2002-02-26 18:21:41
|
Hallo Nate, "Nate" <we...@mo...> wrote: >Oliver, make sure you run it from a console, it prints to stdout if >it gets a regex "compile" error. Was the regex problem a compile >problem or something I did? I can't do much if the regex "library" >has a problem, we just call it and hope it works. The regex seems to work fine. I have problems with the or function. search: dancer dark string: bjork bjoerk /// 1000 10000 filter: or size With ??? I got messages like the following: Filter: or, file='Bjork - Dancer in the Dark - I've Seen It All.mp3', string='bjork bjoerk ' There are a lot of matching songs in the search window, but no one is queued. Even without size filter it doesn't work. ~ Oliver |
|
From: Nate <we...@mo...> - 2002-02-26 06:08:09
|
Sorry about that, with the last CVS update I broke uploads. I updated CVS and it works again. I tested the resume feature too just to make sure. |
|
From: Nate <we...@mo...> - 2002-02-25 23:02:18
|
I know the dup file thing is a pain, I have the same problem and will probably get to it soon. It would be nice to be able to sort your files into different directories, and we could do that via the filter thing too! By the way, I am adding a "and" filter. I realized you could do a "and" after doing another filter and get some specific results. Scanman pointed out several months ago that a file's size will most likely not be the exact same (if it's over say 100,000 bytes) so we can use that as a "marker" for dups. Plus, I could do a SHA1 thing on the first 3 or 10K of a file to see if it's the same. Scanman wrote some code using size and it's at the end of the downloads.c file, but not used yet. The problem is the time it takes to go through a long list of "done" files (say you had 10,000 or more) for every "hit" you get back in a search. That's why it was nice to just use the Linux file system built in b-tree to do it for us. I probably need to do a internal b-tree or something. It's a lot of work and I wanted to get this filter thing out so you could start using it. Anyone got some good well tested easy to use, written in C b-tree code samples? I may think up some stuff you can use till I get something better going, but it will be slow. What I mean by slow is if you have a filter sitting there for 3 seconds doing something, you arn't servicing the network and can get dropped connections. Oliver, make sure you run it from a console, it prints to stdout if it gets a regex "compile" error. Was the regex problem a compile problem or something I did? I can't do much if the regex "library" has a problem, we just call it and hope it works. Once nice thing about filters is if one has a problem, just don't use it, or don't send it that string. At least the whole program isn't trashed by one problem. |
|
From: Oliver T. <ol...@bo...> - 2002-02-25 13:31:18
|
Hallo Antti, Antti Boman <ant...@mi...> wrote: >Nate wrote: >> I updated CVS with the new concept for the automation filters. I just >> barely got it working, and regex is in but lightly tested. If you use >> regex, let me know if you have any problems. Nate, thanks for implementing the regex feature. But the regex has failed at least in one of my tests. I will test it again and give you a short feedback. >> It's pretty easy now to add your own filter. You will have to change >> your automation config file, just rename your old one so napshare >> generates a new one, then copy / paste in your old file settings and >> do a search/replace. Yeah, it is easy and very powerful. >You could pick up any file from the download queue, right click, and >choose "Ignore this file", which would move it (name+size) to a global >exclusion list. The file would never be seen again. Also, one could >configure Napshare to exclude all the successfully downloaded files. The >names of the files could be edited to be regexps. > >Pros: >- You wouldn't have to keep the files in the download directory >- You wouldn't have to write complicated regexps I've sorted my whole napshare download directory into one directory for each interpret. Now napshare is sucking all the downloaded songs again. And often I have the same song with different filenames on my harddisk, but if I don't want to download them again I have to leave them. And so the 'exclude list' is a nice suggestion. ~ Oliver |
|
From: Antti B. <ant...@mi...> - 2002-02-25 10:54:49
|
Nate wrote: > I updated CVS with the new concept for the automation filters. I just > barely got it working, and regex is in but lightly tested. If you use > regex, let me know if you have any problems. > > See the "Automation Filters" I added to the README file, it explains > all. > > It's pretty easy now to add your own filter. You will have to change > your automation config file, just rename your old one so napshare > generates a new one, then copy / paste in your old file settings and > do a search/replace. > > I will be working on cleaning it up a bit, and adding a few more > filters. Bug reports would be great at this point. > > Enjoy! I haven't tested this yet (I will, and thanks for the work), but this reminds me of the possible file exclusion list. So, is there a need for a list of files (regexps with sizes) that should never be shown in searches? You could pick up any file from the download queue, right click, and choose "Ignore this file", which would move it (name+size) to a global exclusion list. The file would never be seen again. Also, one could configure Napshare to exclude all the successfully downloaded files. The names of the files could be edited to be regexps. Pros: - You wouldn't have to keep the files in the download directory - You wouldn't have to write complicated regexps I'm willing to start this up, in a few weeks, in case I get the permission. I won't do it for myself only, and I don't want to maintain another branch of competing Napshare. Just give me a sign ;) -a |
|
From: Nate <we...@mo...> - 2002-02-24 10:15:13
|
I updated CVS with the new concept for the automation filters. I just barely got it working, and regex is in but lightly tested. If you use regex, let me know if you have any problems. See the "Automation Filters" I added to the README file, it explains all. It's pretty easy now to add your own filter. You will have to change your automation config file, just rename your old one so napshare generates a new one, then copy / paste in your old file settings and do a search/replace. I will be working on cleaning it up a bit, and adding a few more filters. Bug reports would be great at this point. Enjoy! |
|
From: Oliver T. <ol...@bo...> - 2002-02-22 19:26:12
|
Hallo Nate, "Nate" <we...@mo...> wrote: >I will have the NOT filter of course, and regex looks possible now >thanks to Oliver, and a size filter (how good I don't know). I am >open to any others! Sure, you have to improve the size filter. I don't want files smaller than 1MB, because this can't be good quality, and I don't want files bigger than 10MB, because these big files are often mpegs. Maybe you add one field for min and one for max size. >The idea is to pass the data to each filter one at a time, so you can >stack filters as you wish (to a limit). Sounds great. ~ Oliver |
|
From: Nate <we...@mo...> - 2002-02-22 04:44:47
|
If you would like to have your "needed" automation filter added to the next release, please post your request in the features section of the SF page. Or, if you make it a subroutine (filename/data in, TRUE or FALSE out) I should be able to get that to work. I thought of a way to allow lots of "advanced" filters and still keep things somewhat simple for beginners. This should be good. I am working on the code now. I will have the NOT filter of course, and regex looks possible now thanks to Oliver, and a size filter (how good I don't know). I am open to any others! The idea is to pass the data to each filter one at a time, so you can stack filters as you wish (to a limit). |
|
From: Oliver T. <ol...@bo...> - 2002-02-20 13:03:25
|
[I wrote this message on monday, but I forgot to cc to the list.] Hallo Antti, Antti Boman <ant...@mi...> wrote: >> Now I can insert a search string (like 'rem') in the search-field and in >> the not-field, I can put a regex (like '^rem[^[:alpha:]]+.*\.mp3$'). > >This requires changing the not-field to mean something else (as here >you're and'ing it with the search string). That's ok though, I'd like to >use the field both ways, so we'll try to come up with something. Of course, often it is easier to use some 'not' words instead of an complex regex. Maybe a "[] use regex" checkbox will do it? >> Maybe someone with higher skills can implement the feature in napshare >> 1.0.1? >What's wrong with your skills and the feature you've made? I've just pasted code from one sourcecode to another. But my success gives me the push to put the K&R "Progamming C" off the bookshelf and restart learning C. ~ Oliver |
|
From: Nate <we...@mo...> - 2002-02-19 20:37:46
|
Gnutella handshake v0.6 is now working. I would consider this beta code though. You can get it via anon CVS. Host connections are a bit faster now and it saves IPs that you are connected to before exit so they are used first when you start up again. It also caches IPs of ultrapeers and X-Try headers when connecting, these come mostly from limewire clients. Enjoy! |
|
From: Antti B. <ant...@mi...> - 2002-02-19 11:12:08
|
Hi, If the disk, where configuration files reside, runs out of space, exiting napshare will delete all the configuration. I'll make a patch that warns the user about it, and doesn't try to save configuration at all. I've ran into this a few times, and it's damn undesirable. This message is also a reminder to myself. More patches to come, when I get time for them. I hope somebody applies them some day. -a |
|
From: Antti B. <ant...@mi...> - 2002-02-18 09:32:48
|
Oliver Tonnhofer wrote: > Hello, > > I started using napshare a few days ago and the automation is great. > But, it has some weaknes. I was searching for songs from R.E.M. and > guess what happens? napshare started to download songs with rem, remind, > remeber, remix, ... in the filename. > > So what I've missed was regex. And today I started to hack a regex > feature into napshare. I'm not a coder, I have just a little C > knowledge. But after less than an hour it works :) > > I used the regex.h from the glibc and code from sylpheed 0.7.0. > > Now I can insert a search string (like 'rem') in the search-field and in > the not-field, I can put a regex (like '^rem[^[:alpha:]]+.*\.mp3$'). This requires changing the not-field to mean something else (as here you're and'ing it with the search string). That's ok though, I'd like to use the field both ways, so we'll try to come up with something. > Maybe someone with higher skills can implement the feature in napshare > 1.0.1? What's wrong with your skills and the feature you've made? -a |
|
From: Oliver T. <ol...@bo...> - 2002-02-16 16:39:05
|
Hello, I started using napshare a few days ago and the automation is great. But, it has some weaknes. I was searching for songs from R.E.M. and guess what happens? napshare started to download songs with rem, remind, remeber, remix, ... in the filename. So what I've missed was regex. And today I started to hack a regex feature into napshare. I'm not a coder, I have just a little C knowledge. But after less than an hour it works :) I used the regex.h from the glibc and code from sylpheed 0.7.0. Now I can insert a search string (like 'rem') in the search-field and in the not-field, I can put a regex (like '^rem[^[:alpha:]]+.*\.mp3$'). Maybe someone with higher skills can implement the feature in napshare 1.0.1? ~ Oliver |
|
From: Nate <we...@mo...> - 2002-02-10 05:37:51
|
Hi Antti, Thanks for your feedback about the program. Please feel free to post and upload patches to the patch manager, see the main web page for instructions (use -u mostly). |
|
From: Antti B. <ant...@mi...> - 2002-02-04 17:21:16
|
Hi, The list doesn't seem to be too active. Anyway, I have thought of (and implemented some) usability improvements. Is there any use sending them here, ie. do you even check them out? I don't mean everything will be included, just wondering if any one of the developers really follows the list. Thanks, agb |
|
From: Antti B. <ant...@mi...> - 2002-01-30 10:57:57
|
ScanMan wrote: > On Tue, 2002-01-29 at 07:44, Antti Boman wrote: >> >Would it help, be "proper" or cause problems if I allowed quotes in >> >the "not" keywords to group them together? Like "music mp3" would mean >> >both the word music and mp3 would have to be in a file name for it to >> >NOT be queued. >> >>That's easily mistaken with a phrase. How about a possibility to use the >>keywords (and or not) and parentheses, with 'or' being the default >>operator? Like "(music and mp3) notmusic" (without quotes ;) > > The way it is now, there are two boxes, one for keywords, and one for > NOT-keywords/ Um, misleading words used. I meant "(music and mp3) or somethingelse", didn't mean to mix the boxes together. So that would match (ie. exclude) files which have both "music" and "mp3", _or_ only "somethingelse". Also, by "and or not" above I meant "and/or/not". >> >Is regex something I should seriously consider? Would you spend the >> >time to support and debug all the regex code? Is it worth it? >> >>Along with keywords/parentheses, this would make the program more >>appealing to hc users. I'd love to see regexes, and I'm willing to help >>with this one, too. > > Yes, tht would be cool. It would also be fairly easy. Which regexp lib should we use then? I tried finding the most common, and at least Debian seems to have libregex (GNU regular expression library) apt-get'able. Is it the one we should start with (if we started)? >> >One thing I want to try not to do is rename a file in the done >> >directory. I am hoping people will just share their done directory so >> >files on the network will be more available. We have enough problems >> >with renamed files now. >> >>Do you mean renaming files already downloaded or files to be downloaded? > > He's talking about partial files. Er... let's try to solve this one (as I didn't get it ;). If you have two different files with the same name, what should the program do? Partial files shouldn't matter, as they aren't (shouldn't be) shared, but the latter of the files with the same name must be renamed if downloaded. Naturally the renaming must occur when the first bytes are written onto the temp dir. >> >What do you think about a option to log operations of automation to a >> >file? It would be put in the same place as the config file. >> >>Sounds good. It could also be a start point for a possibility to discard >>files by name/size (as now such files must be first downloaded and kept, >>even if you don't want them at all). > > That's a REALLY good idea. I'm not the guy in charge; what do you think of this, Nate? More about this: There will be overhead with the checks, so the list should maintain itself. Something like auto-clearing discarded files if they haven't been seen for n days/months at all. Just brainstorming. Sorry if I din't make myself very clear ;) -a |
|
From: Antti B. <ant...@mi...> - 2002-01-29 12:48:13
|
Hi, I'm new to the list, and quite new to the program itself, but still very enthusiastic (especially about the automation part). I'm willing to help with development. But anyway, on with the subject: >Would it help, be "proper" or cause problems if I allowed quotes in >the "not" keywords to group them together? Like "music mp3" would mean >both the word music and mp3 would have to be in a file name for it to >NOT be queued. That's easily mistaken with a phrase. How about a possibility to use the keywords (and or not) and parentheses, with 'or' being the default operator? Like "(music and mp3) notmusic" (without quotes ;) >Is regex something I should seriously consider? Would you spend the >time to support and debug all the regex code? Is it worth it? Along with keywords/parentheses, this would make the program more appealing to hc users. I'd love to see regexes, and I'm willing to help with this one, too. >One thing I want to try not to do is rename a file in the done >directory. I am hoping people will just share their done directory so >files on the network will be more available. We have enough problems >with renamed files now. Do you mean renaming files already downloaded or files to be downloaded? >What do you think about a option to log operations of automation to a >file? It would be put in the same place as the config file. Sounds good. It could also be a start point for a possibility to discard files by name/size (as now such files must be first downloaded and kept, even if you don't want them at all). -a |
|
From: Nate <we...@mo...> - 2002-01-15 15:17:16
|
Hi all, I got tired of testing this code, fixed what bugs were reported, ran out of time for a while so I decided to ship the prototype! Automation is working well, hosts load faster and the host size is limited, file selection works better, memory seems to stay at a reasonable size and other fixes. It's not perfect but it seems to be pretty stable as is. Please try to compile this on whatever platform you have access to and let me know if it worked so I can make a list of what it will run on. Please post any bug reports to the bug manager on SF. Enjoy! http://napshare.sourceforge.net/ |
|
From: M. R. B. <mr...@0x...> - 2002-01-06 06:32:17
|
* Goswin Brederlow <gos...@st...> on Sat, Jan = 05, 2002: >=20 > Implement md5sums for files (md5sum? read the RFC). lopster for > example uses md5sum and when you download a file it compares filesize > and md5sum for matches. It also warns you if you have a file of that > size and md5sum in you shared folder. >=20 It would be silly to md5 hash the entire file once it's already downloaded, better to hash the first 10k or so (or however much can be guaranteed entropy) and use that *and* file size to compare incompletes. The only way hashing the entire file makes sense is if it's already sent as part of the HTTP headers. M. R. |
|
From: Goswin B. <gos...@st...> - 2002-01-05 19:01:52
|
"Nate" <we...@mo...> writes:
> Hi all,
>
> It would be nice to hear everyones opinion on the following:
>
> Evan (scanman857) has pointed out some interesting possibilities for
> selecting files.
>
> He suggests using the file size to find duplicate files. Even allowing
> resume download if a file has the same exact size but a different
> name.
A very crufty solution to a design problem of gnutella that realy
sucks.
Implement md5sums for files (md5sum? read the RFC). lopster for
example uses md5sum and when you download a file it compares filesize
and md5sum for matches. It also warns you if you have a file of that
size and md5sum in you shared folder.
Size alone is just way to insecure.
Apart from that YES.
MfG
Goswin
PS: md5sums would also allow checking the download for correctness.
|
|
From: Nate <we...@mo...> - 2002-01-05 13:47:14
|
Hi all, It would be nice to hear everyones opinion on the following: Evan (scanman857) has pointed out some interesting possibilities for selecting files. He suggests using the file size to find duplicate files. Even allowing resume download if a file has the same exact size but a different name. I was thinking of using the search keywords and/or? the "not" string as a check too, along with the file size, is this needed? Should I make several "filter modes" and let the user select them? I will try to run some tests. When thinking about this you have to consider 650MB files and also 10K txt files people will be downloading. Mp3 files are near the same size, so if you had a large collection of say 3000 files, what about that? How many files in your collection are exactly the same? Would it help, be "proper" or cause problems if I allowed quotes in the "not" keywords to group them together? Like "music mp3" would mean both the word music and mp3 would have to be in a file name for it to NOT be queued. Is regex something I should seriously consider? Would you spend the time to support and debug all the regex code? Is it worth it? Should I request say 3K less on a file resume if available, then watch to see if it "syncs" with the file I am now receiving? If I don't have 3K should I just write over the old file? Would it be worth it to have a "database" of say the first 10K of a "done" file and check against that? or a md5 of the 10K? Is it worth it to add some sort of external program call (perl, tcl or regex) to the check for automation? Even if it slows down the check? How do I decide which file name is the best? Should I care? One thing I want to try not to do is rename a file in the done directory. I am hoping people will just share their done directory so files on the network will be more available. We have enough problems with renamed files now. What do you think about a option to log operations of automation to a file? It would be put in the same place as the config file. Thanks in advance for your input on these. |
|
From: Nate <we...@mo...> - 2002-01-04 06:10:38
|
I'm proud to announce NapShare is now fully automated! The new "automation" feature tries to simulate a person doing searches and selecting files all night long. It tries its best to be nice to the network. I created a new gui screen that lists the search phrases and filter words. It saves them to a file which is read in next start. You can even have NapShare start with automation on. It's available now via anon CVS at: http://sourceforge.net/cvs/?group_id=19407 A few things still don't work and I am still testing. Please post bugs to the bug manager and I will try to get to them before the next public release (soon). If you already have NapShare via CVS, and you run into problems with "cvs update", you should make a back up of your napshare dir and then delete all files in the working dir except the CVS dirs, then do a update and it should bring in all new files for you. Don't forget to do a ./configure again. All compile warnings are gone, and the makefile was corrected. Let me know if you have any problems compiling. See the README file for more information on automation. If anyone can create RPM's for me, it would be appreciated (on the next release). Enjoy! |
|
From: Nate <we...@mo...> - 2001-12-24 21:53:35
|
Hi, I spent a lot of hours over the weekend trying to see if I could find any memory leaks. I found a possible one in uploads.c and fixed some other minor stuff. I did it the "hard way", by looking at all strdups and mallocs and seeing if the free code matched. If someone wants to use dmalloc or mark-malloc on this, please do. I don't have the time to figure out how to do it. In testing gtk-gnutella and NapShare both, I find that they will increase memory usage even if you just sit there and do host connects only (no searches or downloads/uploads). So there is still a problem (I think). Also, after you do a search it increases and never decreases. This bothers me because I have been over and over the search.c code many times and everything looks good to me. I am wondering if maybe there is a call to a system routine to "clean up" memory that I should be doing every so many minutes? Or this leak is in a linked library somewhere and I will never find it. It might be possible there is no leak, and it's just users that let their searches run all night, ending up with millions of hits. On initial host connect (none in list), gtk-gnutella code seems to connect to *all* the external hosts at once. I fixed that. I made changes to NapShare and they are now on CVS. Sourceforge has a nice diff feature if you browse CVS, you can easily see the changes I made. Use anon CVS to get the latest version for testing of course. |
|
From: ScanMan <sc...@sc...> - 2001-08-23 06:36:06
|
Currently, NapShare groups files by their name when downloading. It would be better if it grouped them by size, so all files with exactly the same size would be considered the same file. |