Strangeness in repository

[antoineL]

I spotted a number of strange things in the repository. Perhaps some are OK. I'll list all, in case of...

This file is not what its extension suggests (it is a HTML file, apparently from Sourceforge at rev.2082); committed rev.2086, 2021-07-05:

Code: Select all

trunk/install/icons/gestinux_32x32_test.png

The following files are stored with CRLF line endings (as far as I can tell), but it is not the way the "other" files of the same type are using.
I know current tools usually handle the difference, but sometimes they choke, and it is even more a problem when some files are of the "wrong" kind...

Code: Select all

./languages/fr_FR_countries.txt
./languages/he_IS.ini
./languages/it_IT.ini
./util/languages/it_IT_util.ini

./reports/BalanceSheet/FR_Resultat 2033B.ini
./reports/AccountingExport/FR_Export_FEC.mysql

./util/images/tgdatetimeedit.xpm

./unitaccountingexport.pas
./unitcoeffstock.pas
./unitdeliverybill.pas
./unitdeliverybills.pas
./unitedi.pas
./unitinventory.pas
./unitorder.pas
./unitorders.pas
./unitproductstock.pas
./unitqueries.pas
./unitquery.pas
./unitquote.pas
./unitquotes.pas
./unitsalesdoc.pas
./unitsalesdoccopy.pas
./unitsalesdocs.pas
./unitsalesinvoice.pas
./unitusers.pas
./util/gfileedit.pas
./util/gfolderedit.pas

./reports/BalanceSheet/FR_Bilan 2033A.lrf
./util/lrs/gdatetimeedit.lrs
./util/lrs/gimage.lrs
./util/lrs/gnavbuttons_img.lrs

./unitcustomerpayments.lfm
./unitdatamodule.lfm
./unitdeliverybill.lfm
./unitinventory.lfm
./unitorder.lfm
./unitproductstock.lfm
./unitquery.lfm
./unitquote.lfm
./unitsalesdoccopy.lfm
./unitsalesinvoice.lfm
./unitusers.lfm

The MacOSX .patch file is also stored with CRLF line ends, it is not what I would expect but I do not know how it should be and there are no other reference at hand.

[Update]
Perhaps the util/languages/it_IT_util.ini file is now stored with LF line ends.

[tintinux]

Hi
For the first file trunk/install/icons/gestinux_32x32_test.png I agree that there must have been a mess.
It is corrected, the file is now the same as in 1.5.

For the end of the lines of the other text files, I believe that it depends on the system (Linux/Windows) where they were created.
However it is is not important with Lazarus, I don't know what is "the best" line ending, both are handled.

Best regards

[antoineL]

For the end of the lines of the other text files, I believe that it depends on the system (Linux/Windows) where they were created.

Well, I managed to understand what was happening here.
tl;dr: In fact, the files I listed were having CRLF line endings only on (my) Windows platform. So people running Linux won't see anything special.

It occurs because those files (and only them) are marked with SVN property svn:eol-style:native; with that setting, files are stored on the repository as LF-ended, but when checked out on Windows, they are modified to have on-disk Windows CRLF endings. It has no effect when checking out at a Linux box.

I am still not comfortable with the discrepancies; but I agree that as long as releases are prepared on a Linux box, it should not have a noticeable impact.

[tintinux]

Hi

About svn:eol-style property :

The property svn:eol-style:native is advised to avoid errors during merge operations, in a number of sites

But since I try to avoid merges with Delphi and Lazarus sources (for other reasons), I have never noticed any drawback when it is missing...

If you want you can add the missing property... A local config file can also set this property by default, for new sources.