This calendar is supposed to be edited by admin users, but do not yet
have the complete JavaScript code to do so, thus for now i have made it
read-only.
I just found out that this is a feature introduced in PostgreSQL 10,
back in 2017.
Besides this being the standard way to define an “auto incremental
column” introduced in SQL:2003[0], called “identity columns”, in
PostgreSQL the new syntax has the following pros, according to [1]:
* No need to explicitly grant usage on the generated sequence.
* Can restart the sequence with only the name of the table and column;
no need to know the sequence’s name.
* An identity column has no default, and the sequence is better
“linked” to the table, therefore you can not drop the default value
but leave the sequence around, and, conversely, can not drop the
sequence if the column is still defined.
Due to this, PostgreSQL’s authors recommendation is to use identity
columns instead of serial, unless there is the need for compatibility
with PostgreSQL older than 10[2], which is not our case.
According to PostgreSQL’s documentation[3], the identity column can be
‘GENERATED BY DEFAULT’ or ‘GENERATED ALWAYS’. In the latter case, it is
not possible to give a user-specified value when inserting unless
specifying ‘OVERRIDING SYSTEM VALUE’. I think this would make harder to
write pgTAP tests, and the old behaviour of serial, which is equivalent
to ‘GENERATED BY DEFAULT’, did not bring me any trouble so far.
[0]: https://sigmodrecord.org/publications/sigmodRecord/0403/E.JimAndrew-standard.pdf
[1]: https://www.2ndquadrant.com/en/blog/postgresql-10-identity-columns/
[2]: https://wiki.postgresql.org/wiki/Don't_Do_This#Don.27t_use_serial
[3]: https://www.postgresql.org/docs/15/sql-createtable.html
I do not like confirmation messages: they question user’s actions, and
in general it is better to offer an undo option afterward. However, undo
is harder to implement, and currently i do not have time to do this.
The delete for the session is different because the only repercussion
would be to log in again; the user is not in danger of losing any data
whatsoever.
This one has an input to select the icon. It makes no sense to choose
an icon only by name, thus a <select> is not appropriate, and had to
use a hidden input with a row of button to choose the icon from. This
works now only because there are very few icons; we’ll need to choose
a different approach when there are many more icons.
Since now the icons have to be defined in CSS for both the public and
admin sections, i had to split it into a separate file that both sites
can use. I considered the option to “include” that CSS with m4, like
i do for images in demo.sql, but it made everything too complicated
(e.g., having to call make for each change in the CSS), and decided to
load that CSS in a separate <link>.
We discussed with Oriol how to show these “extra” menu items, as they
can’t be in the horizontal menu we have intended for employees, because
there is not enough horizontal space.
Oriol suggested to move these into the user menu. In fact, the company
settings was already there, which means that i already wanted to do that
from the very beginning, i believe, but i must have forgotten it along
the way…. Or maybe it was because this is where Numerus has the company
settings menu item, too, and i did not see the relation with the rest;
i do not know.
I decided to use a custom attribute for the campsite label in the SVG
because i was a bit wary of reusing ‘id’ for that, specially given that
most labels are number only and XML can not have IDs starting with a
number.
In fact, at least Inkscape and Affinity solve the problem by having an
additional foreign attribute to keep the “group label” in without that
restriction (‘inkscape:label’ and ‘serif:id’, respectively), thus i
thought of doing the same, but with a namespace that i control and be
independent of the design program.
However, it seems that Affinity does not have a way of editing the XML
attributes like Inkscape does[0], thus there is no way of adding or
editing that value from there; i can not ask Oriol to edit the SVG file
in a text editor each time.
We have agreed to reuse the ‘id’ attribute to contain the campsite’s
label by using a specific prefix, that we checked is editable without
issue in Affinity’s UI.
[0]: https://forum.affinity.serif.com/index.php?/topic/24318-xml-data/&do=findComment&comment=115609
I am using the US terms for campground and campsite, that’s why the
relation is called ‘campsite’ instead of ‘pitch’, but i used the wrong
terminology in the SVG map because the customer uses the UK term and
call themselves campsite, so i mixed things.
It is now the campground map and each individual area is a campsite,
as i have been using all along.
I did not remember that the file was loaded as a module, so i have
to export it in order to use from the <script> in media. That <script>
needs to be also a module and explicitly import the function; since it
is already loaded, the browser does not load the file again.
This link is supposed to be for employees, to see the map and check
on campsites’ availability. Currently, it shows the same for employees
and admins, but it will need to change.
I intend to use the same SVG file for customers and employees, so i had
to change Oriol’s design to add a class to layers that are supposed to
be only for customers, like trees. These are hidden in the admin area.
I understood that customers and employees have to click on a campsite to
select it, and then they can book or whatever they need to do to them.
Since customers and employees most certainly will need to have different
listeners on campsites, i decided to add the link with JavaScript. To
do so, i need a custom XML attribute with the campsite’s identifier.
Since i have seen that all campsites have a label, i changed the
“identifier” to the unique combination (company_id, label). The
company_id is there because different companies could have the same
label; i left the campsite_id primary key for foreign constraints.
In this case, as a test, i add an <a> element to the campsite with a
link to edit it; we’ll discuss with Oriol what exactly it needs to do.
However, the original design had the labels in a different layer, that
interfered with the link, as the numbers must be above the path and
the link must wrap the path in order to “inherit” its shape. I had no
other recourse than to move the labels in the same layer as the paths’.
Have to call Dialog.showModal when HTMx loaded the dialog in the DOM,
so had to add a onLoad event listened that checks whether the loaded
element is actually a DIALOG.
Had to restrict the margin: 0 for all elements (*) to exclude dialog,
because the browser sets it to auto, and i did not want to set it again
just because i was too overzealous with my “reset”.
The rest of the CSS is just to have a sticky header and footer, and see
the cancel button, that works as a “close”, all the time.
Finally, i realized that if i add the dialog at the end of the fieldset
and let HTMx inherit its hx-target and hx-swap, i no longer need to set
them in the dialog, as HTMx will always replace the fieldset, and i can
have the dialog side by side the current content of the fieldset, that
it was very confusing seeing it disappear when trying to select a new
media.
The cancel button could now just remove the dialog instead of making the
POST, but in local it makes no difference; we’lls see what happens on
production.
It makes easier to upload new images from the place where we need it,
instead of having to go to the media section each time.
It was a little messy, this one.
First of all, I realized that POSTint to /admin/media/picker to get the
new media field was wrong: i was not asking the server to “accept an
entity”, but only requesting a new HTML value, just like a GET to
/admin/media/upload requests the form to upload a new media, thus here
i should do the same, except i needed the query parameters to change the
field, which is fine—it is actually a different resource, thus a
different URL.
Then, i thought that i could not POST the upload to /admin/media,
because i returned a different HTML —the media field—, so i reused the
recently unused POST to /admin/media/picker to upload that file and
return the HTML for the field. It was wrong, because i was not
requesting the server to put the file as a subordinate of
/admin/media/picker, only /admin/media, but i did not come up with any
other solution.
Since i had two different upload functions now, i created uploadForm’s
Handle method to refactorize the duplicated logic to a single place.
Unfortunately, i did not work as i expected because uploadForm’s and
mediaPicker’s MustRender methods are different, and mediaPicker has to
embed uploadForm to render the form in the picker. That made me change
Handle’s output to a boolean and error in order for the HTTP handler
function know when to render the form with the error messages with the
proper MustRender handler.
However, I saw the opportunity of reusing that Handler method for
editMedia, that was doing mostly the same job, but had to call a
different Validate than uploadForm’s, because editMedia does not require
the uploaded file. That’s when i realized that i could use an interface
and that this interface could be reused not only within media but
throughout the application, and added HandleMultipart in form.
Had to create a different interface for multipart forms because they
need different parameters in Parse that non-multipart form, when i add
that interface, hence had to also change Parse to ParseForm to account
for the difference in signature; not a big deal.
After all that, i realized that i **could** POST to /admin/media in both
cases, because i always return “an HTML entity”, it just happens that
for the media section it is empty with a redirect, and for the picker is
the field. That made the whole Handle method a bit redundant, but i
left it nevertheless, as i find it slightly easier to read the
uploadMedia function now.
This is required if requesting an admin page via HTMx; i do not think it
is done as of now, except for the media picker, that has no layout
because i do not want to update the title.
It made no sense to have a file upload in each form that needs a media,
because to reuse an existing media users would need to upload the exact
same file again; this is very unusual and unfriendly.
A better option is to have a “centralized” media section, where people
can upload files there, and then have a picker to select from there.
Ideally, there would be an upload option in the picker, but i did not
add it yet.
I’ve split the content from the media because i want users to have the
option to update a media, for instance when they need to upload a
reduced or cropped version of the same photo, without an edit they would
need to upload the file as a new media and then update all places where
the old version was used. And i did not want to trouble people that
uploads the same photo twice: without the separate relation, doing so
would throw a constraint error.
I do not believe there is any security problem to have all companies
link their media to the same file, as they were already readable by
everyone and could upload the data from a different company to their
own; in other words, it is not worse than it was now.
This page is more or less similar to home, in terms of database: it
has a carousel and a list of items; in this case, the definition of
campsite services.
As i said early, when adding the home carousel, this carousel has its
own relation and set of functions to manage slides. They are also
duplicated in Go code, but i think i will need to refactor it later to
a carousel package or something like that, because both relations have
the exact same fields and types, so it makes no sense to have twice the
same code.
I already did it with the CSS and JavaScript code, mostly because it was
easier to replace the `.surroundings div` selector with `.carousel`, and
because that way i can have a single template that loads and initializes
Slick.
There is no UI to create or edit service definitions, although there are
the SQL functions, because i have no more time now, and Oriol needs to
check that the style is correct for that page.
This page is “highly stylized”, with a masonry-like grid, that i did not
know how to generate automatically from data defined in PostgreSQL,
therefore with Oriol we agreed to have this one as a static page and
we will see what we can do if the customer asks to be able to change
it.
I was a bit undecided on whether the icons in the bottom part of the
page should be defined in the CSS or with style="" and CSS variables,
like i do for the campsite type in the home page.
At first i thought that it should use CSS variables, mostly for
coherence: if another section of the web does it for its background
image, why no this one. The difference is that the home page is
dynamically created from the database, while this page is static and we
know what icons we need, thus it makes more sense to move it to the
stylesheet file, because then it will be downloaded by user agents that
actually want to use it (e.g., browsers, but not Braille terminals).
I debated with myself whether to create the home_carousel relation or
rather if it would be better to have a single carousel relation for all
pages. However, i thought that it would be actually harder to maintain
a single relation because i would need an additional column to tell one
carrousel from another, and what would that column be? An enum? A
foreign key to another relation? home_carousel carries no such issues.
I was starting to duplicate logic all over the packages, such as the
way to encode media paths or “localization” (l10n) input fields.
Therefore, i refactorized them.
In the case of media path, i added a function that accepts rows of
media, because always need the same columns from the row, and it was
yet another repetition if i needed to pass them all the time. Plus,
these kind of functions can be called as `table.function`, that make
them look like columns from the table; if PostgreSQL implemented virtual
generated columns, i would have used that instead.
I am not sure whether that media_path function can be immutable. An
immutable function is “guaranteed to return the same results given the
same arguments forever”, which would be true if the inputs where the
hash and the original_filename columns, instead of the whole rows, but
i left it as static because i did not know whether PostgreSQL interprets
the “same row but with different values” as a different input. That is,
whether PostgreSQL’s concept of row is the actual tuple or the space
that has a rowid, irrespective of contents; in the latter case, the
function can not be immutable. Just to be in the safe side, i left it
stable.
The home page was starting to grow a bit too much inside the app
package, new that it has its own admin handler, and moved it all to a
separate package.
Debian’s pgtype version is 1.10.0, but pgtype did not add RecordArray
type until 1.11.0, thus could not compile the application with
Debian 12.
I just copied the code from 1.11.0 without any modification besides
adding the reference to the pgtype package to types and functions.
I am not happy with the localization interface for admins, but it is the
easier that i could think of (for me, i guess), with a separate for
each language.
I am not at all proud of the use of RecordArray, but i did not see the
need to create and register a type just to show the translation links.
I might change my mind when i need to add more and more translation
links, but only it the current interface remains, which i am not that
sure at the moment.
There is a big difference between the item that has the submenu for
lodges and languages: languages is a link to the “alternate” version of
the page, while the lodges has no page to link to. Therefore, one is an
anchor while the other is a button, to make a semantic difference, but
both have the exact same appearance here.
Since campsite types need a media, i have to insert also images to the
media relation. The best would be to use PostgreSQL’s
pg_read_binary_file to read the media content from actual files when
inserting the new rows, but the files need to be within the database
cluster directory, or have to use an absolute path when running as a
superuser to read from files outside the cluster directory, which means
that it would depend on the path where i leave the files, that is
different in development that in staging.
To avoid that problem i can simply insert the rows using their base64
strings, with PostgreSQL’s decode. The images are kind of small, but
i was worried that each change in demo.sql would duplicate that data in
git, even if the change is not related to the images, because git stores
the whole file; even if small, soon everything adds up.
I do not care if the _final_ demo.sql is big, as this file is packaged
in a different deb and is only installed in staging, so i’ve chosen to
use m4 to build a single “amalgamated” SQL file from the base .sql
file and the individual image files converted to base64 strings. That
way, each image is individually managed by git and the base .sql file
does not balloon up for each little change.
Changed m4’s quotes to [[ ]] because the default ` ' was interfering
with Intellij’s syntax highlighting.
This is the image that is shown at the home page, and maybe other pages
in the future. We can not use a static file because this image can be
changed by the customer, not us; just like name and description.
I decided to keep the actual media content in the database, but to copy
this file out to the file system the first time it is accessed. This is
because we are going to replicate the database to a public instance that
must show exactly the same image, but the customer will update the image
from the private instance, behind a firewall. We could also synchronize
the folder where they upload the images, the same way we will replicate,
but i thought that i would make the whole thing a little more brittle:
this way if it can replicate the update of the media, it is impossible
to not have its contents; dumping it to a file is to improve subsequent
requests to the same media.
I use the hex representation of the media’s hash as the URL to the
resource, because PostgreSQL’s base64 is not URL save (i.e., it uses
RFC2045’s charset that includes the forward slash[0]), and i did not
feel necessary write a new function just to slightly reduce the URLs’
length.
Before checking if the file exists, i make sure that the given hash is
an hex string, like i do for UUID, otherwise any other check is going
to fail for sure. I moved out hex.Valid function from UUID to check for
valid hex values, but the actual hash check is inside app/media because
i doubt it will be used outside that module.
[0]: https://datatracker.ietf.org/doc/html/rfc2045#section-6.8
jordi fita mas
oriol carbonell pujolàs