Обсуждение: [HACKERS] Re: [QUESTIONS] libpq++
I build the old version (from postgres95-1.09) libpq++ with libpq 6.1. It works good enough (at least in my program unde FreeBSD). May be for backward compatibility it will be useful to include the old libpq++ (e.g. as libpqo++) in the postgres distribution? The reason - old libpq++ is simple and does not use C++ template classes. If somebody is interesting in it I can send the code. - ---------------------------------- Sergey Kosyakov Landau Institute for Theoretical Physics E-Mail: Sergey S. Kosyakov <ks@itp.ac.ru> Date: 26-Jun-97 Time: 16:57:08 - ---------------------------------- ------------------------------
On 26-Jun-97 Leo Shuster wrote: > >The previous implementation of libpq++ was full of bugs, leaked memory >all over the place, and suffered from C style programming. That is >why it had to be rewritten. I would not even recommend brining this >monstrosity back to life. The ANSII string class that is used in >libpq++ in its current reincarnation comes standard with g++ and >compiles well everywhere. But it needs some g++ comand-line options hack (at least for FreeBSD), and it is not described anywhere. Moreover, it is not obvious how to use e.g. PgEnv, which uses "string". Of cause, the new libpq++ looks much more attractive. But it needs documentation or, at least, examples. And there are some difficulties in using gdb with C++ templates. > [ skipped ] - ---------------------------------- Sergey Kosyakov Landau Institute for Theoretical Physics E-Mail: Sergey S. Kosyakov <ks@itp.ac.ru> Date: 26-Jun-97 Time: 17:42:48 - ---------------------------------- ------------------------------
>> I just replace my old postres95 installation with PostgreSQL-6.1. I was
>> surprising by libpq++, which I used intensivly. It is almost completely
>> rewrited but man page contains the description of the old version.
>> Unfortunatelly, now I should rewrite my code for all changed libpq++ names,
>> methods, etc...
>Can someone verify this? Do we need the libpq++ manual page rewritten?
Yes. Yes. :-]
Leo
------------------------------
The previous implementation of libpq++ was full of bugs, leaked memory all over the place, and suffered from C style programming. That is why it had to be rewritten. I would not even recommend brining this monstrosity back to life. The ANSII string class that is used in libpq++ in its current reincarnation comes standard with g++ and compiles well everywhere. It does not add too much complexity or size to the final library. It only makes it more stable and maintanable. I would vote against regressing to the previous form of libpq++. Leo _______________________________________________________________________________ Subject: [HACKERS] Re: [QUESTIONS] libpq++ From: "Sergey S. Kosyakov" <ks%itp.ac.ru@interlockp.lubrizol.com> at LZ- INTERNET Date: 6/26/97 4:57 PM I build the old version (from postgres95-1.09) libpq++ with libpq 6.1. It works good enough (at least in my program unde FreeBSD). May be for backward compatibility it will be useful to include the old libpq++ (e.g. as libpqo++) in the postgres distribution? The reason - old libpq++ is simple and does not use C++ template classes. If somebody is interesting in it I can send the code. - ---------------------------------- Sergey Kosyakov Landau Institute for Theoretical Physics E-Mail: Sergey S. Kosyakov <ks@itp.ac.ru> Date: 26-Jun-97 Time: 16:57:08 - ---------------------------------- ------------------------------