[fpc-pascal] SeekEof issue & 2020 blog post by A Stolyarov

Alexey Torgashin support at uvviewsoft.com
Mon Sep 11 11:08:08 CEST 2023


Translation of blog post http://www.stolyarov.info/node/290 from Ru to 
En.
Sep 6 2020.
Author: Andrey Stolyarov, author of Ru books about programming.

==
I'm trying to convince FPC men to finally fix SeekEof:
https://forum.lazarus.freepascal.org/index.php/topic,51353.0.html

Tellingly, I even sent them a fix. Proven fix, Carl! One strange man has 
already run in (Jonas Maebe). It feels like receiving replies from 
officials. I wonder if everyone is like that, or is there still a 
reasonable person? Or maybe I got into the wrong place, I don't need to 
go to the forum, but immediately to the bugtracker?

UPD: I reported in the bug tracker: 
https://bugs.freepascal.org/view.php?id=37716 ( 
https://gitlab.com/freepascal.org/fpc/source/-/issues/37716 ) Place your 
bets, I'll still have to write in the new edition of the book that the 
numbers from the text stream can be read correctly only 
character-by-character, or they will fix it.

UPD2: I received a response from someone in the bug tracker. I made a 
patch according to their rules. Submit. We stock up on popcorn.

UPD3: Christmas trees, where do such people come from and how is Free 
Pascal still alive with such a team?! And, most importantly, what to do 
now? I'm sorry, I won't be able to fork this monster, it's easier to 
write my own Pascal compiler, but it's not for one year of work.

UPD4: https://bugs.freepascal.org/view.php?id=37833 ( 
https://gitlab.com/freepascal.org/fpc/source/-/issues/37833 )- now we 
will convince them to fix the consequences of their fix, which "fixes" 
what is not broken. By God, if it weren't for the fact that there are 
simply no other implementations of Pascal, I would have spat on these 
comrades a long time ago. Sometimes it is easier for students (and what 
is there sometimes, just almost always) to explain what and how than to 
this stubborn character. This time, it's a little worse that I can't 
provide a ready-made patch - well, I probably can, but there are damn 
changes in git, it will be difficult to isolate among them those that 
this alternatively professional comrade made.
====
In the comments of that post:

Q (reader): Judging by the state of the ticket, has success been 
achieved?
A (Stolyarov): To be honest, it's not clear - you'll have to pull out a 
snapshot and see. I really hope that this Tomas Hajny didn't understand 
what happens when you open a text stream and ended up not being able to 
do what he wanted to do -- cut off buffering for anything that isn't a 
file. If he succeeded, then "the victory has some strange taste" - 
SeekEof seems to have been fixed to us, but in general it has become 
even worse than before. To be honest, I didn't expect to encounter such 
a level of militant ignorance.
UPD: I looked. So yes. I don't even know what's worse - what happened 
before this "fix" or what happened after. It is quite difficult to do 
without the correct SeekEof, but you can - for example, read everything 
character by character and translate numbers with pens. But when 
buffering is turned off for output, it is impossible to turn it on from 
inside the program, no means are provided for this. The devil knows 
what.


More information about the fpc-pascal mailing list