- From: <Irene.Vatton@inrialpes.fr>
- Date: Mon, 03 Jan 2000 18:06:56 +0100
- To: Michal Jaegermann <michal@ellpspace.math.ualberta.ca>
- cc: www-amaya@w3.org
Hi, You're right, I'm going to clean up these errors. In-reply-to: Your message of Tue, 28 Dec 1999 00:39:20 -0500." <199912280539.WAA22854@ellpspace.math.ualberta.ca> > I took a closer look at amaya-2.4. Various issues which I mentioned > in the past are still in the current sources - so I will not repeat > myself. Still the biggest gotcha is that all '...handler.c' sources > from thotlib/image are basically broken as they all include a code > in this style: > > void PngPrint (...., int fd, ....) > { > .... > fprintf((FILE *)fd,.... > .... > } > > Surprise! C does not guarantee that pointer will fit into int and > indeed - this happens to be the case on my Alpha, for example - so if > (FILE *) was passed to a function in 'int' we just lost a significant > chunk of it. May work accidentally somewhere but it does not make the > code correct. As a result (not necesarily of the particular piece of > a code above) if I try to look at some 'screenshot' in a graphic > format I am getting a blank screen and maybe a message like this one " > *** Thot: Irrecoverable error ***[1]+ Done(1)". > > >From "not seen before" department I notice quite a few instances > of something like " ...if (c == _TABULATION_) ... " where 'c' > is 'char', or maybe 'int'. The trouble is that from headers > follows that _TABULATION_ is a pointer to a string so a comparison > is not likely to evaluate to "true". > > Other than that things seem to work (if one discounts occasional > complaints about "unaligned access" or crashes). > > Regards, > Michal > michal@harddata.com > > > > Irene.
Received on Monday, 3 January 2000 12:07:09 UTC