1. FakeThreads-Server
- Posted by jbrown105 at speedymail.org Aug 15, 2002
- 826 views
>From my web site you can get a working version of my fake thread library, which abandoned the mutex/shared memory idea and uses a server to handle shared variable requests. It works, and it is my hope that it will be useful for someone one day. Also I'm working on EParts, a library which is based on RDC, but allows multiple servers and well as multiple clients, and eventually peer-to-peer calling of routines instead of just having clients call one server, as RDC is now. BTW, I am uploading my FakeBind package to my site right now as well. Any comments? jbrown -- http://fastmail.fm - the way email *should* be
2. Re: FakeThreads-Server
- Posted by Kat <gertie at PELL.NET> Aug 15, 2002
- 646 views
On 15 Aug 2002, at 19:17, jbrown105 at speedymail.org wrote: > > >From my web site you can get a working version of my fake thread > library, > which abandoned the mutex/shared memory idea and uses a server to > handle shared > variable requests. It works, and it is my hope that it will be useful > for someone > one day. Also I'm working on EParts, a library which is based on RDC, > but > allows multiple servers and well as multiple clients, and eventually > peer-to-peer > calling of routines instead of just having clients call one server, as > RDC is now. > > BTW, I am uploading my FakeBind package to my site right now as well. > Any comments? About using sockets, yeas, a biggie: do NOT use if you are not behind a firewall, or without password protection and ip verification in your Eu code. Opening a port for anyone to access is like handing your keybd to strangers. The http://www.rapideuphoria.com/mirc.zip shows how to add Eu threads to mirc. The same can add threads to Eu itself. Kat
3. Re: FakeThreads-Server
- Posted by jbrown105 at speedymail.org Aug 17, 2002
- 640 views
On 0, Kat <gertie at PELL.NET> wrote: > > On 15 Aug 2002, at 19:17, jbrown105 at speedymail.org wrote: > > > > > >From my web site you can get a working version of my fake thread > > library, > > which abandoned the mutex/shared memory idea and uses a server to > > handle shared > > variable requests. It works, and it is my hope that it will be useful > > for someone > > one day. Also I'm working on EParts, a library which is based on RDC, > > but > > allows multiple servers and well as multiple clients, and eventually > > peer-to-peer > > calling of routines instead of just having clients call one server, as > > RDC is now. > > > > BTW, I am uploading my FakeBind package to my site right now as well. > > Any comments? > > About using sockets, yeas, a biggie: do NOT use if you are not behind a > firewall, or without password protection and ip verification in your Eu code. > Opening a port for anyone to access is like handing your keybd to strangers. > > The http://www.rapideuphoria.com/mirc.zip shows how to add Eu threads to > mirc. The same can add threads to Eu itself. > > Kat > Can I get it to work for Linux? My FakeThreads-Server package is Linux-only (It may work for FreeBSD tho) while your package looks like its Windows-only (I assume from the presence of Win32Lib). I agree that my method is not secure, but the port which is used is chosen randomly and can not be known ahead of time, and nmap, a portmap scanner, appears not to be able to find the port which my threads server connects to. This makes is unlikely that someone will be able to break in (I agree its not 100%, but it does pose a formidable barrier). Also, any Linux user with a direct connection to the internet will be running some sort of firewall, unless they found some way to connect their system without running any servers. That said, using sockets for a program designed to work locally is a bad idea, but I was stuck with getting mutexes to work (it appeared that multiple programs would access the database anyways an corrupt it). jbrown -- http://fastmail.fm/ - Access your email from home and the web
4. Re: FakeThreads-Server
- Posted by Kat <gertie at PELL.NET> Aug 17, 2002
- 645 views
On 17 Aug 2002, at 18:51, jbrown105 at speedymail.org wrote: > > On 0, Kat <gertie at PELL.NET> wrote: > > > > On 15 Aug 2002, at 19:17, jbrown105 at speedymail.org wrote: > > > > > > > > >From my web site you can get a working version of my fake thread > > > library, > > > which abandoned the mutex/shared memory idea and uses a server to > > > handle shared > > > variable requests. It works, and it is my hope that it will be useful > > > for someone > > > one day. Also I'm working on EParts, a library which is based on RDC, > > > but > > > allows multiple servers and well as multiple clients, and eventually > > > peer-to-peer > > > calling of routines instead of just having clients call one server, as > > > RDC is now. > > > > > > BTW, I am uploading my FakeBind package to my site right now as well. > > > Any comments? > > > > About using sockets, yeas, a biggie: do NOT use if you are not behind a > > firewall, or without password protection and ip verification in your Eu > > code. > > Opening a port for anyone to access is like handing your keybd to strangers. > > > > The http://www.rapideuphoria.com/mirc.zip shows how to add Eu threads to > > mirc. > > The same can add threads to Eu itself. > > > > Kat > > > > Can I get it to work for Linux? My FakeThreads-Server package is > Linux-only > (It may work for FreeBSD tho) while your package looks like its > Windows-only > (I assume from the presence of Win32Lib). Correct, mirc is windows-only, i pleaded with Khaled to make it run on winCE for palmtops, he declined. I hear it runs about half way on Wine on *nix, but the win32lib won't, of course. BUT, you can use the win code on a winbox, and interface to one of the *nix packages (in the archives) on a *nixbox. > I agree that my method is not > secure, > but the port which is used is chosen randomly and can not be known > ahead of time, > and nmap, a portmap scanner, appears not to be able to find the port > which my > threads server connects to. This makes is unlikely that someone will be > able to > break in (I agree its not 100%, but it does pose a formidable barrier). Depending on how you did the nmap, if you did it on yourself, on the same box, it's invalid. The socks layer will short-circuit the probes, and you cannot see yourself from outside the box. The probes won't leave the internet side of the socks layer and re-enter. You may not get a real picture with nmap on a lan either, depending on permissions on the lan router or ip mapper. You should run the nmap from outside your system. > Also, > any Linux user with a direct connection to the internet will be running > some > sort of firewall, unless they found some way to connect their system > without > running any servers. That said, using sockets for a program designed to > work > locally is a bad idea, but I was stuck with getting mutexes to work (it > appeared > that multiple programs would access the database anyways an corrupt > it). Actually, i considered sockets as a great idea, since on a lan you can mix OSs and any type of hardware you can get. Also, you can run several routers, and put a box outside the protected zone, accessable to the internet for serving or fetching, and a beowulf on the inside, protected. The disadvantage to DDE or shared memory is that is cannot be shared with other puters, altho it's very secure on that one puter from interference of another puter. With sockets, you code protection into the socks interface you write, giving read/write permission as needed. I was at one time, years ago, trying to get a team together to run a dedicated box on my isp's lan. The idea was to put a box there running nix, and a dedicated line back here, where i could add any scripting and database and puter array that i wanted to. I won't bore you with the number and ways i was let down, or the amount of opportunities to do non-normal webhosting that passed me by since then. Kat
5. Re: FakeThreads-Server
- Posted by jbrown105 at speedymail.org Aug 22, 2002
- 637 views
On 0, Kat <gertie at PELL.NET> wrote: > > On 17 Aug 2002, at 18:51, jbrown105 at speedymail.org wrote: > > > > > On 0, Kat <gertie at PELL.NET> wrote: > > > > > > On 15 Aug 2002, at 19:17, jbrown105 at speedymail.org wrote: > > > > > > > > > > > >From my web site you can get a working version of my fake thread > > > > library, > > > > which abandoned the mutex/shared memory idea and uses a server to > > > > handle shared > > > > variable requests. It works, and it is my hope that it will be useful > > > > for someone > > > > one day. Also I'm working on EParts, a library which is based on RDC, > > > > but > > > > allows multiple servers and well as multiple clients, and eventually > > > > peer-to-peer > > > > calling of routines instead of just having clients call one server, as > > > > RDC is now. > > > > > > > > BTW, I am uploading my FakeBind package to my site right now as well. > > > > Any comments? > > > > > > About using sockets, yeas, a biggie: do NOT use if you are not behind a > > > firewall, or without password protection and ip verification in your Eu > > > code. > > > Opening a port for anyone to access is like handing your keybd to > > > strangers. > > > > > > The http://www.rapideuphoria.com/mirc.zip shows how to add Eu threads to > > > mirc. > > > The same can add threads to Eu itself. > > > > > > Kat > > > > > > > Can I get it to work for Linux? My FakeThreads-Server package is > > Linux-only > > (It may work for FreeBSD tho) while your package looks like its > > Windows-only > > (I assume from the presence of Win32Lib). > > Correct, mirc is windows-only, i pleaded with Khaled to make it run on winCE > for palmtops, he declined. I hear it runs about half way on Wine on *nix, but > the win32lib won't, of course. BUT, you can use the win code on a winbox, > and interface to one of the *nix packages (in the archives) on a *nixbox. > I'd rather not, as FakeThreads is designed to run threads on the same computer (which is why inet sockets are a bad idea.) I'm looking into redoing RDC.e with UNIX local sockets, but I'm not sure if that will give adequate protection. (I think local UNIX sockets (AF_UNIX) can be used by remote UNIX machines, just as inet sockets (AF_INET) can be used by remote machines, but I'm not entirely sure.) > > I agree that my method is not > > secure, > > but the port which is used is chosen randomly and can not be known > > ahead of time, > > and nmap, a portmap scanner, appears not to be able to find the port > > which my > > threads server connects to. This makes is unlikely that someone will be > > able to > > break in (I agree its not 100%, but it does pose a formidable barrier). > > Depending on how you did the nmap, if you did it on yourself, on the same > box, it's invalid. The socks layer will short-circuit the probes, and you > cannot > see yourself from outside the box. The probes won't leave the internet side of > > the socks layer and re-enter. You may not get a real picture with nmap on a > lan either, depending on permissions on the lan router or ip mapper. You > should run the nmap from outside your system. > Well I use a firewall, so almost all of my ports are blocked from the outside. Of course, its alwayss possible to bypass a firewall in theory. > > Also, > > any Linux user with a direct connection to the internet will be running > > some > > sort of firewall, unless they found some way to connect their system > > without > > running any servers. That said, using sockets for a program designed to > > work > > locally is a bad idea, but I was stuck with getting mutexes to work (it > > appeared > > that multiple programs would access the database anyways an corrupt > > it). > > Actually, i considered sockets as a great idea, since on a lan you can mix > OSs and any type of hardware you can get. Also, you can run several > routers, and put a box outside the protected zone, accessable to the internet > for serving or fetching, and a beowulf on the inside, protected. The > disadvantage to DDE or shared memory is that is cannot be shared with > other puters, altho it's very secure on that one puter from interference of > another puter. With sockets, you code protection into the socks interface > you write, giving read/write permission as needed. > Interesting idea, I may try out a RemoteThreads package designed to allow programs to share "threads" over networks. > I was at one time, years ago, trying to get a team together to run a dedicated > > box on my isp's lan. The idea was to put a box there running nix, and a > dedicated line back here, where i could add any scripting and database and > puter array that i wanted to. I won't bore you with the number and ways i was > let down, or the amount of opportunities to do non-normal webhosting that > passed me by since then. > And I shall not bore you by showing my sympathy, but I will say that you had yet-another-very-clever-idea-which-was-shot-down-for-most-likely- ridiculous-reasons. > Kat > jbrown -- http://fastmail.fm/ - Consolidate POP email and Hotmail in one place