2016-01-25 20:42 GMT-02:00 Ingo Schwarze <schwa...@usta.de>: > Hi Rodrigo, > > redirected to misc@, this is off-topic on tech@. >
OK, my fault > > Rodrigo Mosconi wrote on Mon, Jan 25, 2016 at 06:54:32PM -0200: > > > I would like to receive some help/mentoring. I`m cursing a master degree > > course at PUC-Rio, and I need to "create a useful program that performs a > > service of interest to anyone other than exsively the student." So I > would > > like to create something to the openbsd project. > > > > I would like to know in which areas need some work? I have partial time > to > > work on it and only this half year to do it. Some of the old google > summer > > of code is still need? There are some need to a new daemon, or replace > an > > old one, aligned with the openbsd style ( configuration files, privsep, > > plegde)? > > For a beginner, it's usually easier to do bugfixing in high-qualitive > existing code than to do full-scale auditing or refactoring in > low-quality existing code, and for a beginner, it is even harder > to write new high-quality code from scratch. So to anyone considering > similar projects, make sure that you start regularly finding bugs > and sending patches to fix them at least one year before you want > to do a project that involves writing a new program for OpenBSD > from scratch. Otherwise, it's gonna be tough. > > Besides, we have a hard time advising people what they could do if > we don't know what they are capable of. "Master student at a > university" doesn't mean a lot. I only said about the curse, because simple patchs aren`t allow to this discipline. I must develop the specification, the project, write the code, document it, implement tests, write user documentation. > There are people who don't study > anything, suddenly pop up out of nowhere, and do good work right > away. There are people with a degree who fail even at simple tasks. > Rodrigo, according to the ChangeLog, you worked on cwm(1) and > cfengine so far. Given that amount of public information, the only > person who can judge your skills is yourself. > > I also worked with zeromq/czmq upstream. In my last job I worked as developer and made some company products > We sometimes say "/usr/bin/" is full, meaning that adding additional > programs doesn't necessarily make the system better. If done without > good reason, it just makes the system larger and less simple. > Consequently, finding a project of the kind "write a new program" > is about ten times harder than finding a project of the kind "improve > an existing program", even for an experienced developer. > > All that said, there _are_ some programs that need to be rewritten. > The one that is most sorely missing is ppp(8), a PPP client program. > That may not be a simple task, and it may not be easy to find a > mentor. The people most knowledgable in that area live in Japan, > they are very focussed and very hard-working people, and they are > *very* busy. > > There is other stuff that requires rewriting, in alphabetical order: > > * adduser(8)/useradd(8): > Needs to be unified into one single program and cleaned up. > re-enginered programs are allowed as subject. > * dig(8)/nslookup(8)/host(8): > A simpler replacement not using external libraries would be useful. > re-enginered programs are allowed as subject. * ftp(8) is very old and suffering from bitrot; the last attempt to > rewrite it trickled out with no obvious conclusion, i don't > exactly know why. > ftp client or ftp server? > * ldapd(8) is decent code, but sorely lacks a maintainer. > re-enginered programs are allowed as subject. > * ldappasswd(1) would be nice to have. > "simple" but usefull > * lpd(8)/lpc(8)/lpr(1) is very old and suffering from bitrot. > I didn`t used lp?, so I skip to work on these > * traceroute(8) needs to be extended by the functionality of > tcptraceroute. > > Some GSOC projects may still be relevant, but one has to admit that > it is rare that GSOC projects produce code that actually gets > committed. Even among those that succeeded, only a minority produced > code good enough to actually get used. If people look for a project > rather than simply working on what they are interested in, chances > for success are quite slim. > I`m looking a interesting project. I will ask to my mentor if the useradd or dns clients can be used as subject. The ldapd probabily is. > That said, don't despair, but you really need to be able to > realistically judge your own skills and interests, make up your own > mind, and take initiative. > I will look for the ldap family first. As user, I have a need for then: the "server" and the "client" part. > > Yours, > Ingo
